Zhida
Li†
a,
Dandan
Yuan†
a,
Hongjun
Wu
*a,
Wei
Li
b and
Di
Gu
a
aProvincial Key Laboratory of Oil & Gas Chemical Technology, College of Chemistry & Chemical Engineering, Northeast Petroleum University, Daqing, 163318, China. E-mail: hjwu@nepu.edu.cn; Tel: +86-0459-6503331
bCollege of Petroleum Engineering, Northeast Petroleum University, Daqing, 163318, China
First published on 6th November 2017
The process of molten salt CO2 capture and electrochemical conversion provides us with a new way to close the present carbon cycle and mitigate global climate change by transforming the greenhouse gas CO2 into carbonaceous fuels or chemicals. In this paper, carbon spheres and carbon nanotubes that can be used as a societal resource to serve mankind are synthesized from CO2 in diverse electrolyte composites with inexpensive metallic electrodes. Carbon products, subsequent to electrolysis, are characterized by EDS, SEM, TEM, Raman, TGA, FTIR, BET and XRD to reveal the elemental composition and morphological and structural features. The results demonstrate that Li–Ca–Na and Li–Ca–K carbonate electrolytes favor carbon sphere formation rather than carbon nanotube formation, and in particular, K2CO3 shows enhanced interference with carbon nanotube growth. In contrast, Li–Ca–Ba and Li–Ba carbonate composites present an increase in the carbon nanotube fraction. Additionally, CNTs generated from Li–K, Li–Ba and Li–Ca–Ba present a different diameter. In this way, the CO2-derived carbon products of carbon spheres and carbon nanotubes could be alternatively synthesized through the appropriate regulation of the electrolyte composition.
Combined with the research hotspot of CO2 removal and global climate change, recent research interest in molten carbonates is increasingly directed to advanced applications in capturing and converting unwanted CO2 into valued carbon materials23–26 and even alkane or syngas on co-electrolysis with hydroxide.27–29 During this electrochemical process, electrons directly act as strong redox reagents to reduce the target compounds and finally achieve the transition from CO2 to harmless, value-added carbonaceous chemicals. Other reported CO2 reduction pathways, such as photochemical and biological methods, demonstrate a satisfactory CO2 degradation performance as well, but their scale-up applications are limited by their complexity and high cost of operation.30 In comparison, the molten salt based electrochemical method provides a more facile route to close the carbon cycle and alleviate the intensifying greenhouse effect. However, most of the CO2-derived carbon products using the indicated method are amorphous due to a lack of further investigation and optimization of the electrolysis conditions.31–33 More recently, Licht's group introduced the carbon nanofiber (CNF) synthesis in pure molten lithium carbonate, where metallic nucleation sites redeposited on the cathodic surface played a critical role in the CNF growth process.34 In this paper, rather than expensive Li2CO3, we present several lower cost carbonate composites, involving binary carbonates and ternary carbonates, to split CO2 into structured materials of CSs and CNTs. To the best of our knowledge, this is the first attempt to synthesize CSs from CO2via molten salt electrolysis. This molten salt electrolyzer, from the carbon cycle and energy storage points of view, provides an alternative route for humans to resolve global excessive CO2 emissions and, simultaneously, finishes the energy transition and storage from conventional electricity to chemical energy.
The elemental composition of the cathodically generated carbon product is revealed by using an energy dispersive spectrometer (EDS, X-Max, British Oxford Instruments Co., Ltd). The micromorphology and microstructure of these electrolytic carbon products are investigated by scanning electron microscopy (SEM, SIGMA, Zeiss Company) and transmission electron microscopy (TEM, JEM-2100, JEOL). Raman microspectroscopy (Raman, DXR, Thermo Fisher Scientific) and Fourier Transform Infrared Spectroscopy (FTIR, Tensor 27, Bruker) are used to reveal relevant structural features. Thermogravimetric analysis (TGA, Diamond, PerkinElmer Co., Ltd) is utilized to study the thermal stabilities. The Brunauer–Emmett–Teller (BET, Tristar-3020, Micromeritics Instrument Corporation) method is employed to determine the surface area. Furthermore, to study the crystal form of the carbon materials with spherical and tubular structures, carbon products deposited in Li–Ca–Na, Li–Ca–K, Li–Ca–Ba and Li–Ba are characterized by X-ray diffraction (XRD, D/MAX 2200, Rigaku).
Fig. 1 Thermodynamic stability analysis of Li2CO3 under diverse conditions.36 |
This interaction was further experimentally verified when foreign Li2O was added into molten Li2CO3. There was no mass loss found in Li2CO3–Li2O (90:10, wt%) at 850 °C for 5 hours. Moreover, with a larger fraction of Li2O (82% Li2CO3–18% Li2O), a mass gain was even observed after 5 hours duration. Besides Li2O, CaO can also serve as an absorbent for CO2 capture, but Li2O is much more dissolvable in eutectic carbonates. As further presented by Wang's group, a Li2O-containing ternary Li–Na–K carbonate mixture also exhibited a visibly enhanced CO2 dissolution process.39 All this information suggests that molten carbonates can readily achieve the regeneration process under a CO2 atmosphere or with sufficient Li2O introduction.
Li2CO3 → Li2O + CO2 | (1) |
CO2 + Li2O → Li2CO3. | (2) |
Additionally, fused carbonates, taking molten Li2CO3 as an example, have 10 moles of tetravalent carbon per liter available for subsequent electrochemical reduction; in contrast, the CO2 concentration in air is only ∼0.04%, which is equivalent to 10−5 moles of tetravalent carbon per liter.26,28,40 Thus, compared to atmospheric CO2, molten carbonates demonstrate a million-fold increase of reducible tetravalent carbon sites per unit volume, which could accelerate the carbon deposition reactions.
Relevant electrochemical reactions, involving carbon deposition at the cathode, oxygen formation at the anode, the regeneration process of carbonate melts, and the overall reaction in this electrolyzer, can be described with the following equations.25
Cathode reaction:
CO32− + 4e− → C + 3O2−. | (3) |
Anode reaction:
2O2−–4e− → O2. | (4) |
Electrolyte reproduction:
O2− + CO2 → CO32−. | (5) |
Overall reaction:
CO2 → C + O2. | (6) |
As shown by Fig. 2a, a thick layer of carbon products aggregates on the galvanized iron cathode due to the direct splitting of CO32− during the reaction process (eqn (3)). Some frozen carbonates are found to adhere to the carbon surface and therefore acid leaching is shown to be necessary. Fig. 2b shows the purified carbon products after a series of post-treatments, and we used EDS to analyze the elemental composition within the indicated detection region (SEM image, right bottom), as displayed in Fig. 2c. C occupies an overwhelming molar ratio of over 90% with O and Au taking up the remaining less than 10%. Doubtlessly, Au comes from the coating operation for SEM-EDS characterization. As to the presence of O in carbon materials, Tang considers that the cathodically generated products are likely to be elemental carbon with few oxygen containing functional groups.41 The oxygen-containing groups are common for active carbons and are believed to be preferable for capacitance performance.42,43
Fig. 2 (a) Photograph of the cathode after electrolysis; (b) purified carbon powders; (c) EDS analysis of the obtained carbon products. |
The production capacity of a molten carbonate electrolyzer has been analyzed based on Faraday's law,25 and generally, 1 A h generates 0.1 g carbon materials and consumes approximately 0.4 g CO2. Now, we are trying to scale up the electrolysis setup to 10 A (cathode area 50 cm2, anode area 100 cm2), where 24 g carbon materials are prepared with 96 g CO2 consumed per day.
NiO + 2e− → Ni + O2−. | (7) |
Fig. 3 depicts the SEM images of the carbon product formed in Li–Ca–Na (66.7:20:13.3, wt%) and Li–Ca–K (66.7:20:13.3, wt%) carbonate electrolysis between a galvanized iron cathode and a nickel anode at 750 °C. Evidently, the observed carbon conformation is seen to consist of nano-scale CSs that take up a dominant fraction of ∼90% in the cathodically deposited carbon products in both Li–Ca–Na and Li–Ca–K carbonates. Compared with Li–Ca–K carbonate melts (Fig. 3c and d), CSs from the Li–Ca–Na system (Fig. 3a and b) exhibit a smaller size, only one half or even less than that generated from the Li–Ca–K system. During the electrolysis process, implementation of a low initial current density is evidenced by the recorded electrolysis voltage of ∼1 V at a current density of 10 mA cm−2, as shown in Fig. 4. This low current density step sustains continuous formation of nickel metal deposits at the cathode surface that subsequently results in structured carbon growth. Different from the previously reported pure Li2CO3 electrolysis with a nickel anode,26 in Li–Ca–Na and Li–Ca–K carbonate melts, metal nucleation sites promote a multidimensional carbon deposition, which leads to CSs rather than CNTs being the preferable structure. As to the applications of CSs, excellent thermal stability, unique electronic properties, and most importantly the tailored structure enable CSs to be applied in different technical fields, such as lithium batteries, as a catalyst support in drug delivery, encapsulation of active transition metals, etc.22 Besides, as previously reported, after being heated to high temperatures, the CSs become graphitic and exhibit an increasing capacity behaviour of 280 mA h g−1 compared with 220 mA h g−1 for normal spheres.45 Further investigations on the application of the prepared CSs are expected in future studies.
Fig. 3 Carbon products generated from ternary Li–Ca–Na and Li–Ca–K carbonate melts at 750 °C (a and b are SEM images of the Li–Ca–Na electrolyte; c and d are SEM images of the Li–Ca–K electrolyte). |
We have noted that Na2CO3 and K2CO3 are more corrosive to the nickel anode than molten Li2CO3, and in this manner, nickel release and redeposition as nickel nucleation sites at the cathode in Na/K-containing electrolytes is theoretically preferable to pure Li2CO3. However, as proved in Fig. 3, CSs instead of CNTs become the primary structure following Li–Ca–Na and Li–Ca–K molten carbonate electrolysis, indicative of visible interference from Na+, K+ or Ca2+ in CNT structure formation. However, as far as we know, Li–Ca carbonates also yield a large number of CNTs,40 and therefore the CS growth may be attributed to the presence of Na2CO3 and K2CO3. The potential needed by Na+ and K+ for reduction to the corresponding alkali metals may provide answers to the question how CNT growth is restrained and CS growth is facilitated. Compared with Li2CO3, the lower reduction potential required by Na2CO3 to deposit Na metal and K2CO3 to deposit K metal makes Na and K metals co-deposit on the cathode as one reduction product which is thermodynamically preferred. As calculated from the relevant thermodynamic data taken from the website of NIST,46 the reduction potential needed by Na2CO3 and K2CO3 to precipitate Na and K metals at 750 °C falls in the range of 2.2–2.4 V, lower than the initial cell voltage of more than 2.5 V during electrolysis at 200 mA cm−2. These data provide supporting evidence for the conjecture that carbon production is accompanied by alkali metal deposition. In fact, in the last century when the nature of the primary cathodic reaction in molten electrolytes was not definitely established, it had been widely suggested that the cathode process included alkali metal deposition and carbon generation possibly due to a series of secondary reactions.47 On the basis of the above analysis, we speculate that the CS structure is formed in Li–Ca–Na and Li–Ca–K eutectic carbonates as follows: carbon growth starts at nickel nucleation sites, but for later stages, the co-deposited alkali metal, coupled with the interface modification of CaO from CaCO3 splitting, interferes in the one-dimensional carbon growth to form CNTs and facilitates multi-dimensional carbon growth to generate the CS structure. In addition, the cell–voltage curves plotted in Li–Ca–Na melts and Li–Ca–K melts present a similar trend when an increasing current density is applied. In particular, the voltage of 200 mA cm−2 decreases with extended duration as the galvanized iron cathode gradually evolves into a graphite electrode with a larger active surface area (carbon deposits cover the original cathodic surface).
Besides the primary generation of CSs, bits of CNTs could still form at the cathode subsequent to Li–Ca–Na carbonate electrolysis, as proved in Fig. 5. Tangled CNTs intersperse among the CS structure, taking up an approximate fraction of 10%, illustrated by SEM images of low magnification (Fig. 5a, b and c). Fig. 5d presents a clear and magnified CNT structure, characterized by a diameter of 100–200 nm. Nickel nanoparticles originating from the low initial activating current step nucleate the CNT structure formation. It is found that except for useful nanoparticles located at CNT tips, there are still some nickel nanoparticles aside from the nanotube structure, useless for CNT formation and growth.34 Exclusion of Na2CO3 from Li–Ca–Na carbonate melts (i.e. eutectic Li–Ca carbonates), interestingly, provides a co-production of CNTs and cement (CaO).40 As further reported, CNTs synthesized from Li–Ca carbonate melts (with sufficient CaCO3 addition, 50 g Li2CO3/10 g CaCO3) possess a thin-walled structure, ∼50 nm wall thickness, compared with 200 nm wall thickness of CNTs derived from a pure Li2CO3 electrolyte.40 Thereby, selective synthesis of CNT and CS structures could be alternatively controlled by excluding/adding Na2CO3 and K2CO3, where the interference from Na+ and K+ seems to be favourable for CS structure formation.
As with electrolysis in Li–Ca–Na and Li–Ca–K melts, prior to high carbon growth currents, a low initial current is applied to form nickel nucleation sites. Fig. 6a and b show lower-magnification SEM images of a mixture of CNTs and CSs, while Fig. 6c and d show the micromorphologies of CNTs and CSs, respectively. Obviously, a yield of ∼50% but less than 80% CNTs (marked with orange arrows) with a diameter of ∼200 nm or larger is presented, comixing with clusters of CSs (marked with green arrows) and irregular carbon particles. It should be noted that the electrolysis is performed in a 30 mL nickel crucible with the inner wall acting as the anode rather than using nickel wire as the anode. This measure can effectively avoid anode fracture, often occurring at the interface between the electrolyte and the atmosphere, due to the severe corrosion caused by high temperature and the molten salt environment.
Fig. 8 (a–c) TEM images of CNTs and CSs; (d) Raman spectrum of CNTs and CSs; (e) FTIR spectrum of CNTs and CSs; (f) TGA of CNTs and CSs. |
The Raman spectrum is recorded to evaluate the graphitization degree of the synthesized CSs and CNTs, as depicted in Fig. 8d. The Raman peaks arising from 1350 cm−1 and 1580 cm−1 correspond to the characteristic disorder-induced mode (D band) and the high frequency E2g first order mode (G band), respectively. The intensity ratio between the D band and G band (ID/IG) is cited as a significant parameter to evaluate graphitization. For CNTs of the Li–Ba system, the ID/IG value falls in the range of 0.7 to 0.8, well consistent with the commercial CNTs, while the ID/IG value of CSs formed in the Li–Ca–K system increases to ∼1.1, indicating higher surface disordering caused by the K+-induced interference and the CaO-induced interface modification.
Fig. 8e shows the FTIR spectrum of the CNTs (Li–Ba system) and CSs (Li–Ca–K system), which demonstrates a good conformance. Peaks at 3450 cm−1 and 1110 cm−1 correspond to the stretching vibration of –OH and C–C–O, respectively. The presence of the C–C–O group provides supporting evidence for the conclusion that O element of the EDS spectrum is related to the oxygen containing functional groups. Moreover, the FTIR peak at 1600 cm−1 signifies the unsaturation property of the CNTs and CSs, which may endow carbon materials with an excellent conductivity, but further relevant measurements are required to verify this assumption.
CNTs (Li–Ba system) and CSs (Li–Ca–K system) are further tested by TGA (air, 10 °C min−1) measurement to analyse their thermal oxidation temperature, as depicted in Fig. 8f. Initially, the minor weight loss below 350 °C might be due to the release of volatile matter, such as moisture. Then, compared with the mass loss temperature at approximately 350 °C of CSs, CNTs demonstrate a strengthened thermal stability of over 400 °C, evident in the top and bottom panels of Fig. 8f, respectively. An additional observation is that after being heated to 650 °C, almost nothing is left, suggesting that the carbon materials are carbon plus a few oxygen functional groups without any other impurities.
The surface area of the carbon structures synthesized by the indicated method, as we have previously noted, significantly depends on the applied temperature and cell voltage/current density.25,26 Generally, lower temperature and larger cell voltage/current density favour a higher BET surface area. Herein, the measured BET surface areas of CSs synthesized in Li–Ca–K carbonates and a mixture of CNTs and CSs deposited in Li–Ca–Ba carbonates are 18 and 11 m2 g−1, respectively. The low surface area value can be attributed to the higher applied temperature of 750 °C, which accelerates the carbon deposition and facilitates compact carbon structure formation. Moreover, according to a previous report, the surface area of CSs varies widely from less than 2 m2 g−1 to very high values, over 1200 m2 g−1, and hard spheres are associated with low surface area carbons, signifying that the CSs synthesized in this work are mainly hard spheres.22
To further reveal the structural features of these carbon materials, X-ray diffraction is employed to characterize their crystal structure. Graphene,50 fullerenes51 and CNTs52 are sp2 carbon bonded materials, and in particular, CNTs have an excellent laminar hexagonal graphite crystalline structure.53 X-ray diffraction patterns of the carbon product prepared in the Li–Ca–Na, Li–Ca–K, Li–Ca–Ba and Li–Ba carbonate systems are depicted in Fig. 9a–d.
For carbon products obtained from Li–Ca–Na and Li–Ca–K carbonates, a graphite peak (26°) with a large half-peak width is found, as evident in Fig. 9a and b, implying a messy arrangement of atoms and a lower graphite level. These random atoms may be related to the interference of Na+ and K+ that subsequently results in an uncontrolled, multidimensional growth orientation. In addition, compared with Li–Ca–K, carbon materials from Li–Ca–Na electrolysis confirm a stronger crystal degree as proved by the apparently enhanced peak intensity. This finding suggests that the CNT fraction in the Li–Ca–Na system is a bit larger than that in the Li–Ca–K system. It has been demonstrated that exclusion of Na2CO3 and K2CO3 from electrolytes endows the corresponding carbon material with an increased CNT fraction. The sharp and clear carbon characteristic peak at 26° (002) in the XRD patterns of Li–Ca–Ba and Li–Ba carbonates (shown in Fig. 9c and d) can be assigned to the hexagonal graphite and diffraction planes, and represents an improved crystallinity compared with that of Na/K-containing carbonates. Additionally, resolved XRD peaks at 43° and 44° (marked with a red arrow in Fig. 9c and d) are observed and provide supporting evidence for the homogeneity of the electrolytic CNTs formed in Li–Ca–Ba and Li–Ba composites.
Footnote |
† These authors contributed equally to this research work. |
This journal is © the Partner Organisations 2018 |