Adrian
Lindner
,
Hannes
Radinger
*,
Frieder
Scheiba
and
Helmut
Ehrenberg
Institute for Applied Materials, Karlsruhe Institute of Technology, 76344 Eggenstein-Leopoldshafen, Germany. E-mail: hannes.radinger@kit.edu
First published on 11th May 2022
Thermal activation of graphite felts has proven to be a valuable technique for electrodes in vanadium flow batteries to improve their sluggish reaction kinetics. In the underlying work, a novel approach is presented to describe the morphological, microstructural, and chemical changes that occur as a result of the activation process. All surface properties were monitored at different stages of thermal activation and correlated with the electrocatalytic activity. The subsequently developed model consists of a combined ablation and damaging process observed by Raman spectroscopy, X-ray photoelectron spectroscopy and scanning electron microscopy. Initially, the outermost layer of adventitious carbon is removed and sp2 layers of graphite are damaged in the oxidative atmosphere, which enhances the electrocatalytic activity by introducing small pores with sharp edges. In later stages, the concentration of reaction sites does not increase further, but the defect geometry changes significantly, leading to lower activity. This new perspective on thermal activation allows several correlations between structural and functional properties of graphite for the vanadium redox couple, describing the importance of structural defects over surface chemistry.
Thermal treatment of GF at 400 °C for 30 h was already proposed in the early 1990s, and the parameters have not changed significantly over the years.8 The higher cell efficiencies were initially attributed to an increased concentration of surface oxygen. Two influencing factors have been the subject of further investigation: treatment time and activation temperature. The use of shorter treatment times at higher temperatures such as 500 °C for 4 h appeared to have similar effects on GF, leading to an increase in active surface area and cell efficiency.9 A reduction in overvoltages was observed after treatment in an oxygen-enriched atmosphere, whereas activation without oxygen in the atmosphere showed no significant improvements.10 While these conclusions are still relevant, other interpretations have since been made. It has been suggested that an increase in surface area and roughness, due to damaging oxidation of GF, is responsible for the increased activity instead of chemical surface functionalization.9,11,12
X-ray photoelectron spectroscopy (XPS) is used to study the chemical composition of an electrode surface. In addition to quantifying the type and concentration of oxygen groups, it is possible to distinguish between sp2 and sp3 hybridized carbon. It has been shown that both, differently bound oxygen and carbon hybridization, affect the reaction kinetics.13,14 Surface oxygen, while most probably not an active reaction site, increases the polarity for better electrode wettability and the surface area accessible to the electrolyte.15 Raman spectroscopy allows an in-depth study of the structural properties of carbon-based materials.16–18 It is possible to characterize not only the degree of disorder but also the type of defects.19,20 Studies on model electrodes have shown that edge sites allow faster electron transfer for vanadium redox reactions.21 For GF, it was subsequently demonstrated that defect density is related to electrocatalytic activity.22–24 Recent computational and experimental approaches suggest that oxidation can be considered as a by-product of lattice defects, suggesting that future studies should focus on non-oxidative activation techniques.25,26
A thorough understanding of how a modification technique alters the surface properties of an electrode can be considered key to a mechanistic understanding of chemical reactions. Finding correlations between the number of lattice defects, chemical composition, and electrochemical performance will enable this approach. Therefore, we have studied in detail the interplay between electrocatalytic activity and physiochemical properties for thermally activated GF. Scanning electron microscopy (SEM) and Raman spectroscopy were used to study defects at the micro- and nanoscale in relation to oxidation time. The chemical composition was monitored by XPS. A detailed comparison of all parameters allowed us to propose a model for thermal activation, and sheds light on the important properties to effectively use GF as an efficient electrochemical energy converter. Furthermore, we show how the above-mentioned analytical tools can be used in combination to develop a holistic view of the structural properties of graphite.
Raman spectroscopy was performed using a Horiba Scientific LabRAM HR Evolution equipped with a 633 nm laser and a 100× magnification objective, resulting in a spot size of ∼2 µm. High wavelength and low laser power are required for high resolution and to avoid material damage. Peak intensity ratios are used to determine various features, using Gauss–Lorentz profiles after spline background correction. ID/IG and ID/ID′ ratios were used to evaluate defect density and features.
The morphology of the samples was studied by scanning electron microscopy (SEM, Merlin, Carl Zeiss) at an accelerating voltage of 5 kV and a probe current of 150 to 500 pA.
The chemical composition was analyzed by XPS using a K-alpha+ spectrometer (Thermo Fisher Scientific) with monochromatic Al-Kα radiation (Ephoton = 1486.6 eV) and a spot size of ∼400 µm. Survey spectra were recorded with a pass energy of 200 eV, and detail spectra with 50 eV. A Shirley background correction was applied and individual species were deconvoluted by Gauss–Lorentzian peak profiles. The asymmetry of the sp2 hybridized carbon was evaluated with a tail mix of 90% and a tail exponent of 1. The positions of the residual components in the C 1s region were constrained to the position of the sp2 carbon, and the FWHM (full width at half maximum) values were constrained to sp3 carbon with a tolerance of ±0.1 eV for each. The O 1s region was deconvoluted by several peaks spaced 1 ± 0.1 eV apart, with the FWHM constrained by ±0.1 eV.
Two different CV measurements were performed: first, in a non-faradaic potential window with increasing scan rate from 10 to 250 mV s−1 to calculate the electrochemical double layer capacitance (EDLC). Second, between 0.2 to 1.6 V for the positive, and −0.05 V to −0.85 V for the negative half-cell response with scan rates from 1 to 10 mV s−1. The CV curves were iR-corrected using electrochemical impedance spectroscopy (EIS) at the OCV. EIS was performed in a frequency range from 1 Hz to 100 kHz. At an applied potential of 0.9 V and −0.450 V, the charge transfer resistance (RCT) of each vanadium redox reaction was evaluated. The impedance spectra were deconvoluted using the RelaxIS 3 software (rhd instruments). All measurements were described by the same equivalent circuit, using a resistor R for the junction and a RQ element for the electrode–electrolyte interface.
The microstructural changes caused by the activation were visualized by SEM. Untreated GF (Fig. 1a) showed a smooth surface, except for the characteristic stripes parallel to the fiber length resulting from the fabrication process. Activation at 400 °C led to the formation of numerous small pore defects after 8 h (Fig. 1b). At longer heating times of 16 and 24 h (Fig. 1c and d), both the size and the number of defects present increased. After 32 h (Fig. 1e), the individual pores were still growing in size, but their edges appear less sharp than at shorter heating times. This trend became even more pronounced after 40 h (Fig. 1f), showing highly concentrated but dull defects.
Fig. 1 High-resolution SEM images of GF. A single fiber is displayed for (a) pristine and (b–f) thermally activated GF after (b) 8 h, (c) 16 h, (d) 24 h, (e) 32 h, and (f) 40 h at 400 °C. |
Raman spectra were recorded for a more detailed characterization of the induced damages. The characteristic signals are the defect-induced bands D (∼1330 cm−1) and D′ (∼1620 cm−1), and the graphite-related band G at ∼1580 cm−1 (Fig. 2a). The ID/IG ratio is typically used to determine the degree of disorder in carbonaceous material, while the ID/ID′ ratio provides information on the type of the defect. Due to the heterogeneity of GF, the activation was repeated several times to inspect multiple charges. Consistent with SEM observations, the number of defects increased during thermal activation up to 16 h, as evidenced by the increase in the ratio of D to G from ∼2.00 to ∼2.25 (Fig. 2b). Longer heating times did not lead to more disorder. The ID/ID′ ratio decreased from ∼3.5 for untreated GF, with a clear minimum between 8 h and 16 h. At longer durations, the ratio increased again. An intensity ratio of ∼3.5 is related to the presence of edge defects, consistent with the observation that the well-defined pores exhibit dull shapes after long activation times. Lower ratios originate from several edges being stacked in the direction perpendicular to the sheet.27
Oxygen concentration and sp2, sp3, and carbon vacancy (CV) content were assessed by XPS. The oxygen and carbon regions were deconvoluted to determine the content of individual surface groups (Fig. 2c and d). CO (∼531.3 eV), C–OH (∼532.3 eV), and COOH (∼533.3 eV) groups were identified in the O 1s spectra. The fraction of graphitic (∼248.4 eV) and amorphous carbon (∼285.1 eV) was determined from the C 1s spectra. Ion bombardment of graphite in vacuum was in other studies used to associate another compound below sp2 carbon at ∼284 eV with carbon vacancies.28 In this work, all spectra were deconvoluted using this additional feature to investigate correlations to microstructural analysis by SEM and Raman spectroscopy.
All surface groups were quantified to study the evolution of the chemical composition as a function of the duration of thermal activation (Fig. 2e). The surface of untreated GF comprises about 69 at% sp2 carbon and 15 at% oxygen. In addition, sp3 carbon accounted for about 16 at%, and no vacancies were detected. The most noticeable change was an increase in CV to ∼13 at% after 16 h of thermal treatment. Longer heating times of up to 40 h resulted in a subsequent decrease in vacancies to ∼10 at%. Similarly, the total oxygen content peaked after 16 h (∼22 at%). Further activation initially led to a decrease to ∼8 at% after 32 h, but then to a renewed increase to ∼18 at% after 40 h. The concentration of sp3 carbon showed no appreciable change during the first 16 h, but after 24 h there was a notable decrease and subsequent increase in sp2 carbon.
In Fig. 3c, ΔEp is plotted versus time to show that the catalytic activity for the VIII/VII redox reaction is almost unchanged after the first few hours of thermal activation. A similar trend was observed in the evaluation of charge transfer resistances (RCT) by EIS (Fig. 3d). For the VIII/VII reaction, the RCT decreased from 6 Ω for GF to about 1 Ω after more than 24 h (Fig. S3†). Considering the VVO2+/VIVO2+ redox couple, the high RCT of GF (25 Ω) was reduced to about 5.5 Ω after 16 h (Fig. S4†). The results obtained from CV and EIS show a different evolving behavior over the thermal activation period depending on the half-cell reaction. A more pronounced evolution over time is observed for the positive half-cell reaction. The two activity parameters discussed (ΔEp and RCT) both show a dramatic decrease at low activation times between 4 and 12 h, followed by a subsequent increase. This corresponds to an improved electrocatalytic activity for short activation times, which is reversed by overlong heating. Over the entire period studied, both ΔEp and RCT are higher for the VVO2+/VIVO2+ than for the negative half-cell reaction. Moreover, the ip ratios for the VIII/VII redox pair are closer to ideal reversibility (=1) over the entire temperature range, while for the positive half-cell the reversibility improves in the early stages of thermal activation but decreases again after more than 16 h (Fig. S5†).
Based on our experimental observations, we propose a model for the thermal activation process that shows three important stages of GF (Fig. 4). On the outside, the felt electrodes consist of several, mostly sp2 hybridized graphite layers. The outermost layer is covered by adventitious carbon, which forms during prolonged atmospheric contact. When pristine GF is introduced into an oxidizing atmosphere, the adventitious carbon is removed and the outermost layer of intact graphite is damaged, leading to the formation of defects that are saturated by oxygen-containing groups. The defects formed in the initial phase of activation are small and have sharp edges, as determined by SEM and Raman spectroscopy. With increasing treatment time, they initially grow due to oxidation of sp3 hybridized carbon, which was quantified by XPS. However, the defect size is finite due to the slower oxidation of sp2 carbon under the applied conditions. This means that after an activation time of about 24 to 32 h, most defects have reached their final shape, which is confirmed by the electrochemical data showing no changes in RCT after 24 h of treatment.
We conjecture the following mechanism for the thermal activation of GF: initially, small and sharp defects are formed at sites of structural weakness due to a higher local concentration of sp3 carbon. With increasing treatment time, more and more sp3 carbon is oxidized, leading to the growth of defects and a morphological change from sharp to blunt edges. As a result, the concentration of the active reaction sites declines, leading to increased RCT and overall lower performance of the positive half-cell reaction. When the defects reach a certain critical size, all locally available weaker sp3 carbon is oxidized and defect growth stops. Only a higher temperature or other changes in atmospheric conditions could further increase the defect size at this point.
Fig. 5 Structure–activity correlation of thermally activated GF. Correlation between (a) RCT and defect type, (b) RCT and CV, (c) RCT and oxygen concentration, (d) CV and defect type. |
According to several studies on the activity of carbon-based electrodes for VFBs, the oxygen concentration at the surface increases the electrochemical performance, especially for the negative half-cell. The underlying work does not allow a conclusive statement on the functionality of oxygen due to lack of correlations. Recent studies suggest that the functional groups of oxygen are overestimated, and most activation processes that generate oxygen at surface defects are overestimated for the charge transfer process, since activation procedures that create oxygen functional groups also increase other kind of defects like carbon vacancies.5 Fig. 5c and S7† show the RCT and ΔEp in relation to surface oxygen content. Neither is the VIII/VII reaction affected by the number of oxygen groups, nor can a correlation between the activity of for the positive half-cell reaction and the presence of oxygen groups be established.
This complements our previous studies on that topic and once again showcases that oxygen has no positive effect on the catalytic activity of GF.23,26 We advocate a new approach to the concept of active reaction sites on graphite for vanadium redox reactions: instead of considering oxygen groups on the surface as active species, more efforts should be made to understand the quality and type of edge sites and carbon vacancies. According to our results, the positive half-cell reaction is the bottleneck when thermal treatment is used as an activation technique. Both the overvoltage and cell resistance were significantly higher. The ability of GF to catalyze the VVO2+/VIVO2+ redox couple depended strongly on the microstructural properties of the electrode. In comparison, the activity for the negative half-cell was largely unaffected by changes in microstructure and chemical composition.
Because of the very different effects of thermal activation on the activity of the vanadium redox reactions, the measurements were performed in half-cell configuration. Experiments in a full-cell introduce a new set of performance-limiting factors that make it impossible to distinguish between the influences, as shown in this work. The goal of this work was not to determine the best performing electrode, but to clarify the fundamental effects of thermal treatment.
In addition, SEM, Raman spectroscopy and XPS can be used complementarily to analyze the structural properties of graphite. These analytical tools evaluate the integrity of the carbon lattice and provide multiple insights into the defect composition from the micro to the nanoscale. The electrochemical activity correlated very well with the visual appearance of the pores in the SEM images, the ID/IG and ID/ID′ ratio determined by Raman spectroscopy, and the CV concentration in XPS. Fig. 5d shows how the CV content and the ID/ID′ ratio are related. For thermally activated GF, a higher number of vacancies corresponded to a lower D to D′ ratio.
Footnote |
† Electronic supplementary information (ESI) available. See https://doi.org/10.1039/d2ra02368g |
This journal is © The Royal Society of Chemistry 2022 |