lable at ScienceDirect

Carbon 106 (2016) 208e217

Contents lists avai

Carbon

journal homepage: www.elsevier .com/locate /carbon

One-pot synthesis of nanostructured carbon materials from carbondioxide via electrolysis in molten carbonate salts

Hongjun Wu a, Zhida Li a, Deqiang Ji a, Yue Liu a, Lili Li a, Dandan Yuan a,Zhonghai Zhang b, Jiawen Ren c, Matthew Lefler c, Baohui Wang a, **, Stuart Licht c, *

a Provincial Key Laboratory of Oil & Gas Chemical Technology, College of Chemistry & Chemical Eng., Northeast Petroleum University, Daqing, 163318, Chinab Department of Chemistry, East China Normal University, 500 Dongchuan Road, Shanghai, 200241, Chinac Dept. of Chemistry, George Washington University, Washington, DC, 20052, USA

a r t i c l e i n f o

Article history:Received 17 March 2016Received in revised form8 May 2016Accepted 10 May 2016Available online 14 May 2016

* Corresponding author.** Corresponding author.

E-mail addresses: wangbh@nepu.edu.cn (B. Wang

http://dx.doi.org/10.1016/j.carbon.2016.05.0310008-6223/© 2016 The Authors. Published by Elsevier

a b s t r a c t

As the primary culprit of greenhouse effect, carbon dioxide has garnered global attention, and thetechnologies currently being developed to reduce the emission of CO2 vary widely. In this study, CO2 waselectrochemically reduced in various molten mixtures of LieNaeK carbonates to carbon nanomaterials.By regulating the electrolysis current density, electrolyte, and electrolytic temperature, the carbonproducts had different morphologies of honeycomb-like and nanotubular structures. A transition from ahoneycomb/platelet to nanomaterial carbon morphology was observed to occur at ~600 �C with increasein temperature. The observation of nanostructures is consistent with a higher diversity of structurespossible with enhanced rearrangement kinetics that can occur at higher temperature. A high yield of acarbon nanotube (CNT) was not observed from a LieNaeK electrolyte, no CNTs are formed from a NaeKcarbonate electrolyte, but a high yield is observed from pure Li, or mixed LieNa or mixed LieBa car-bonate electrolytes, and the carbon nanotube product diameter is observed to increase with increasingelectrolysis time.© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

In recent years, there have seen a series of explorations into thesequestration and utilization of CO2 due to problems involvingenvironmental deterioration, increasing global energy demand, aswell as the growing enthusiasm alongside the expansion of sus-tainable energy utilization. However, despite our efforts, most ofthe renewable energy sources and traditional fossil fuels still lead tosignificant CO2 emissions. Therefore, it is of vital importance tocapture CO2 and to transform it into energetic vectors or fuels, suchas carbon, carbon monoxide (CO), or even methane [1]. CO2 is ahighly stable, noncombustible molecule, and its kinetic stabilitypreviously had made its transformation into an economic, stable,non-greenhouse material a challenge [2]. Different strategies forCO2 splitting have been proposed, and the quantity of research inthis field has grown rapidly [3e10].

Recent studies on photocatalysis usingmetal complexes [11e25]

), slicht@gwu.edu (S. Licht).

Ltd. This is an open access article u

and metal oxide based photoelectrodes [26e29] demonstrate CO2reduction performance. However, the catalytic activities and se-lectivities of these photocatalysts have not yet achieved their op-timum performance. Over the years, a new CO2 reduction strategyof high temperature electrolysis has been proposed and receivedsignificant attention. As an example, Dipu et al. investigated theelectrochemical conversion of CO2 into CO in a molten salt elec-trolyte at cathode of YSZ (yttria stabilized zirconia) doped withLa0.8Sr0.2MnO3 [30]. As another example, the continuous conver-sion of CO2 into CO by electrolysis at 900 �C has been studied, andcurrent densities higher than 100 mA/cm2 have obtained [2,31].Besides conversion to CO, CO2 the electrolytic reduction of car-bonates to solid carbon in inorganic molten electrolytes (specif-ically from hydroxides and a barium chloride/barium carbonatemelt) was recognized as early as the late 1800’s [32], and continuesto be of interest today [2,33e35] including the excellent work of theChen group [36e38] andWang groups [39]. The previous examplesof electrochemical conversion processes require the consumptionof electrical energy, and the next challenging task we face is tofigure out how to supply that energy using sustainable energysources. In 2002, a theory was developed for the high efficiency

nder the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

H. Wu et al. / Carbon 106 (2016) 208e217 209

conversion of solar to chemical energy by using the full spectrum ofsunlight [40,41]. This includes, both photovoltaic’s super-bandgapand solar thermal’s sub-band gap energy to drive high tempera-ture electrolyses. This theory was experimentally demonstratedusing a small bandgap (1.1 eV, Si) semiconductor for high solarefficiency solar water splitting in molten salts in 2003 [42]. ThisSTEP (solar thermal electrochemical process) theory was general-ized to the synthesis of useful chemical staples in 2009 [43], andexperimentally demonstrated with several new chemistries[44e48]. With solar power, our team has achieved the electro-reduction of CO2 into energetic materials in molten melts[2,34,47,49].

Most recently, we have demonstrated the unexpected highyield, low energy synthesis of carbon nanofibers and carbonnanotubes (CNTs) from regular and isotopic lithium carbonate[32,50,51]. Here, we explore the one-pot synthesis of nano-structured carbon materials from several other electrolytes. In thisnew study, as summarized in Fig. 1, we observe that the diameter ofCNTs grown from Li2CO3 (mp 723 �C) increases with growth time,and for a lower melting LieNaeK carbonate eutectic (mp 399 �C),there is a transition from a honeycomb/platelet to nanomaterialcarbon morphology is be demonstrated to occur at ~600 �C withincrease in temperature. We also show that the yield of CNTs is lowin the LieNaeK carbonate eutectic electrolyte at all temperaturesinvestigate, and that CNTs are not formed by electrolysis from a Na/K carbonate electrolyte. However, as with pure Li2CO3, a high yieldof CNT products is formed at the cathode using Li/Na and Li/Bamixed carbonate electrolytes.

Fig. 1. One pot electrolytic syntheses demonstrated in this study show for the cathode prodnew results show longer electrolyses (20 h right vs 2 h left) yield larger diameter CNTs. Mnanostructures become evident at T > 600 �C, but CNTs remain <15% of total C even at 770 �CHowever, Li/Na (middle) and Li/Ba (right) carbonate electrolytes each can generate a high

2. Experimental

2.1. Preparation of carbon products

The electrolyte, Li2CO3eNa2CO3eK2CO3 (61:22:17 wt%, analyti-cally pure), was added into an alumina crucible and heated to750 �C. A NieCr spiral (f1.6 mm, 9.9 cm in length, 5 cm2, HebeiSteady Metal Products Co., LTD, China) and a Fe spiral (f1.6 mm,9.9 cm in length, 5 cm2, Hebei Steady Metal Products Co., LTD,China) were used as the anode and cathode, respectively. The areais determined from the cathode wire p � diameter � length.Electrode current densities are determined by this simple, constantmacroscopic area, and do not include microscopic morphologyareas (such as area changes as carbon is deposited on the cathodeduring the electrodeposition). To avoid the effect caused by oxidefilm on the electrode surface, the electrode was polished beforeelectrolysis and stored in oil. The alumina crucible was kept at750 �C for several hours to ensure that the eutectic carbonatemixture was molten and all moisture had been removed. Thetemperature of the alumina crucible was then cooled down to thetemperature required for the electrolysis.

The electrolysis was performed at current densities of 200 mA/cm2 and 400 mA/cm2 at temperatures of 577 �C, 602 �C, 627 �C,652 �C, and 677 �C in eutectic carbonates. The total electrolysischarge was kept constant during the experiments, that is when thecurrent density was 200 mA/cm2, electrolysis time was 4 h, andwhen the current density was 400 mA/cm2, electrolysis lasted for2 h.

After electrolysis for the times indicated above, the carbonproducts electro-deposited at the cathode were removed from the

uct: Top: Li2CO3 (mp 723 �C) electrolysis can form carbon nanotubes at high yield andiddle: Left to right: LieNeK carbonate At T < 600 �C forms platlet products. Carbon. Bottom: In only Na/K carbonates (without Li) no significant carbon product is formed.CNT product yield. (A colour version of this figure can be viewed online.)

H. Wu et al. / Carbon 106 (2016) 208e217210

electrolytic cell, and were subject to a series of post-electrolysistreatments. The cathodic products were rinsed with diluted hy-drochloric acid to remove excess electrolyte and other impurities(e.g. oxides), and then they were placed into a 60 �C water bath for~2e3 h to remove the residual dilute hydrochloric acid. Afterwards,the products were filtered with 0.4 mmwater film and washed withultrapure water until no chlorine ions were detected. Finally, theproducts were dried in a cylindrical drying oven for 4 h at 105 �C.

2.2. Characterization

The composition of obtained products was characterized byEnergy Dispersive Spectroscope (EDS, X-Max, the British OxfordInstruments co., LTD), and Thermogravimetric Analysis and Dif-ferential Thermal Analysis (TG/DTA, Diamond, the United StatesPerkin Elmer co., LTD). The morphologies and structures of thecarbon were studied by Raman Microspectroscopy (RAM, LabRAMHR800), Brunauer-Emmett-Teller analysis (BET, Tristar-3020,Micromeritics Instrument Corporation), Laser particle analyzer(Nano-ZS90, Malvern), Raman Microspectroscopy (RAM, LabRAMHR800), Phenom Scanning Electron Microscope (SEM with Pro-XEnergy Dispersive Spectroscopy), SIGMA SEM e Zeiss company),FEI Teneo LV SEM, and Transmission Electron Microscope (TEM,Tecnai/G2/F20/S-TWIN7, FEI company in the United States).

3. Results and discussion

3.1. Composition analysis of cathodic products

The digital photos of the electrodes before and after electrolysisin a eutectic Li2CO3eNa2CO3eK2CO3 (61:22:17 wt%) electrolyte for4 Ah at 577 �C are presented in Fig. 2a. As can be seen, a thickdeposit of black product with a rough surface is obtained at the Fecathode, covered by small amount of white frozen carbonate

Fig. 2. (a): Photos of the cathode and anode before and after electrolysis. (b): Energy dispersRaman measurement of the products formed at the Fe cathode at a current density of 200

electrolyte. In sharp contrast, except for some slight corrosion andfrozen electrolyte, no other significant changes happen to theanode.

In molten eutectic carbonate the observed electrochemicalprocesses are consistent with Eqs. (1)e(3) equations below [42].CO2 is indirectly converted into solid carbon via reduction of CO3

2�

to elemental carbon and O2� ions in a eutectic carbonate mixture(Eq. (1)); the O2� migrates towards the NieCr anodewhere the ionscan be oxidized to O2 according to Eq. (2). Besides oxidation to O2,the excess O2� formed during the electrolysis captures (combineswith) atmospheric CO2 from the air above the molten carbonateelectrolyte allowing the regeneration of the molten salt electrolyte,Eq. (3), and thus forming a regenerative process of carbon dioxidecapture and carbonate renewal:O2� þ CO2

� / CO32� / C þ O2 þ O2�. These three basic reactions

are, in reality, a simplification of the overall reaction, which in-cludes many other complex reactions that take place in the elec-trolyzer [42].

CO32�(molten-miscible) þ 4e� / C(solid) þ 3O2�(molten-

miscible) (1)

2O2�(molten-miscible) � 4e� / O2(gas) (2)

CO2(gas) þ O2�(molten-miscible) / CO32�(molten-miscible) (3)

Energy dispersive spectrum (EDS) was used to evaluate theelementary composition of the products obtained at indicatedelectrolytic conditions. As can be seen from Fig. 2b, the productsmainly contain two elements: C and O (the peak of Au comes fromthe coating for SEM analysis, and Cl peaks are a result of the dilutedhydrochloric acid used in purification). Molten electrolyte con-taining carbonate and oxide tenaciously holds onto the extractedcathode. Hence, in the absence of the extended product wash

ive spectrum of the cathodic products. (c): TG/DTA curves of the cathodic products. (d):mA/cm2. (A colour version of this figure can be viewed online.)

H. Wu et al. / Carbon 106 (2016) 208e217 211

described in the Experimental section of the cathode, the carbonproduct exhibits EDS without oxygen. With washing, the carbonoccupies the larger mole fraction, reaching up to 95% of the totalproduct, while the mole percentage of oxygen is only 5%, suggest-ing that the products deposited at cathode are likely to beelemental carbon with a few oxygen containing functional groups[39].

TG/DTA measurement was conducted to analyze the thermaloxidation temperature of the carbon products (Fig. 2c). The emer-gence of themass loss at around 320 �C and the exothermic peak onthe DTA curve at nearly 400 �C provide strong supporting evidencefor the identification of the cathodic products, which is in goodagreement with TG/DTA curves of carbon [52]. And finally almostnothing was left, indicating the product was carbon plus someoxygen groups without any other impurities. Both the EDS and TG/DTA characterization are suggestive of excellent generation ofelemental carbon.

To further reveal the internal characteristics of the carbonproducts, their Raman spectrumwas measured and evaluated. Twopeaks are observed at around 1350 and 1580 cm�1 in Ramanspectrum (Fig. 2d), corresponding to the disorder peak (D band)and graphitization peak (G band), respectively. The strength ratiobetween the disorder peak and graphitization peak (ID/IG) is anindicator to assess the graphitization of the carbon products, andthe higher the ID/IG ratio, the stronger the disorder degree will be[53]. In this case, the ration of ID/IG is ~1:1, signifying that the car-bon products are mostly amorphous.

Fig. 3. SEM images of carbon materials obtained at current density of 200 mA/cm2: (a) 577prepared at 200 mA/cm2, 627 �C. (A colour version of this figure can be viewed online.)

3.2. Morphology and structure of the carbon products

Aside from amorphous carbon, special structures in the carbondeposits were also observed. Fig. 3 presents the SEM images of thespecial structures in carbon products obtained at current density of200 mA/cm2 and at different electrolytic temperatures. At thesurface of the carbon powders deposited at 577 �C (Fig. 3a), hon-eycomb shapes manifest with apertures of about 200e400 nm.Different from the nano-sheets and nano-flakes obtained by otherresearchers [39], this honeycomb-like structure is composed ofclosely packed carbon sheets, rather than a scattered distribution.Also shown in Fig. 3a, aside from the honeycomb-like structure,platelet structures with thicknesses ranging from dozens to hun-dreds of nanometers are observed as well, scattered on the surfaceof the honeycomb-like structure.

When the temperature was raised up to 602 �C, the morphologyand structure of carbon materials changed significantly. It is clearthat the honeycomb-like structure at 577 �C begins to be replacedby separate micron size particles that are composed of a variety ofcarbon nanoparticles with diverse diameters and morphologiesthat combine together (Fig. 3b). Furthermore, there is a smallamount of thin tubular structures that intertwine irregularly in theproducts. When the electrolytic temperature was continuouslyincreased, the size and morphology of the carbon nanostructuresfurther changed. As shown in Fig. 3cee, the tubular structures getmore defined with increasing temperature, and it becomes a largerportion of the deposit, though a mixture of carbon nanoparticlesand platelet structure always exists in the products no matter how

�C, (b) 602 �C, (c) 627 �C, (d) 652 �C, (e) 677 �C, (f) TEM image of the tubular structure

H. Wu et al. / Carbon 106 (2016) 208e217212

the electrolytic temperatures change. TEM measurement of thecarbon products strongly proves successive formation of hollowCNT with high commercial application value, shown in Fig. 3f(evidence that these carbon nanostructures are hollow is denotedin the figure), endowing this work with broader prospects. Today’sconventional chemical vapor deposition or electrospun productionvalue is ~$25,000 per tonne of CNF [54] and that of industrial grade(90% purity) CNTs is ~$200,000 to 400,000 per metric tonne [55].Unlike the eutectic LiNaK carbonate, pure lithium carbonate onlymelts 723 �C. When electrolyzed at 750e770 �C, the pure lithiumcarbonate can produce a high yield (in excess of 80%) of uniformCNT and high current efficiency [32,50]. However, although theLiNaK carbonate electrolyte generates a diverse range of carbonnanomaterials at 750e770 �C (not shown in Fig. 3), the yield of anyuniform carbon nanomaterials is very low (less than 15%). Asdescribed in a subsequent section, this low yield LiNaK challenge isovercome by exclusion of potassium from the electrolyte.

Fig. 4 presents the SEM images of carbon products obtained atan electrolysis current density of 400 mA/cm2 and at differentelectrolytic temperatures. When the electrolytic temperature was577 �C, just as with the carbon products at 200 mA/cm2, honey-comb shapes were observed. Similar to the carbon products pre-pared at 200 mA/cm2, the honeycomb morphology of the productsdeposited at 400 mA/cm2 gradually evolves into tubular structureat elevated temperatures, shown in Fig. 4bee. Despite the fact thatthe change of current density has no obvious effect on themorphology of the carbon products, a subtle difference still existsbetween the carbon products prepared at current densities of

Fig. 4. SEM images of carbon materials obtained at current density of 400 mA/cm2: (a) 577prepared at 400 mA/cm2, 627 �C. (A colour version of this figure can be viewed online.)

200 mA/cm2 and 400 mA/cm2. Specifically, the apertures of thehoneycomb-like structures, diameters of the carbon nanoparticlesand nanotubular structures, as well as thickness of the plateletstructures are noted to be different when comparing carbon de-posits produced at 200 mA/cm2 and 400 mA/cm2. A TEM image ofthe tubular structure is exhibited in Fig. 4f, and it can be observedthat CNT still exists in products prepared at 400 mA/cm2, 627 �C,whereas the apertures of the CNT are lower than that prepared at200mA/cm2, 627 �C. Analysis of the SEM images and TEM images inFig. 3 and Fig. 4, diameters of the tubular structures range widelyfrom 20 to 400 nm, and are strongly affected by electrolytic con-ditions. In the process of scanning the samples using SEM, a specialhelical structure (displayed in the top right hand corner of Fig. 4c)was observed. The helical structure looks more neat and orderly incomparison to the irregular intertwined CNT that is the norm,therefore particular attention will be paid to the generation of thespecial helical structure as well as reaction mechanisms involved infuture study.

The generation of these interesting nanostructures could beattributed to the rapid deposition of the carbon nanoparticles[32,50]. The above results demonstrate that both electrolytic tem-perature and current density affect the morphology and structureof the deposited carbon products. Our previous work [56] sug-gested the absolute value of the electrochemical potential of alkalimetal carbonates decreased at elevated electrolytic temperatures,indicating that higher electrolytic temperature could accelerate thedeposition of carbon nanoparticles and favor growth of larger anddenser carbon products. The effect of current density on the

�C, (b) 602 �C, (c) 627 �C, (d) 652 �C, (e) 677 �C, (f) TEM image of the tubular structure

H. Wu et al. / Carbon 106 (2016) 208e217 213

morphology of the products is mainly embodied in the change ofnanostructures’ size, detailed in the paragraph above.

As summarized in Table 1, in the Li2CO3eNa2CO3eK2CO3eutectic, lower temperature synthesis favors a higher surface areacarbon product, and this correlates with the observed averageparticle size variation with temperature in Fig. 4. Note, that thisparticle size is an estimate based on the observed SEM andcorroborated by Nano-ZS90 Laser particle analysis, and does notinclude refinements for the observed irregular particle shapes andplatelet thicknesses. The measured BET surface area for the carbonproduct synthesized at 200 mA cm�2 (not included in Table 1), was10 m2/g at 677 �C, and this is similar to that measured for carbonsynthesized at 400 mA cm�2.

3.3. Current efficiency of the carbon products at differentelectrolysis system

It is observed fromVet curves (Fig. 5a and b) that voltages of theelectrolysis system decrease with elevated electrolytic temperatureat constant current density. Although several competing reactionsapply [35], in general the electrolytic potential for carbonatesplitting thermodynamically decreases with increasing tempera-ture. Also consistent with the observed decrease in electrolyticpotential is that both electrolyte resistance decreases and the ki-netics of charge transfer are facilitated with increasing tempera-ture. Using 200 mA/cm2 as an example, the voltages values are~2.48V at 577 �C and ~2.26V at 677 �C, respectively. Meanwhile, bycomparing voltages at 200 mA/cm2 and at 400 mA/cm2, it can befound that voltages increasewhen applied current density increase,in accord with the expected increase in overpotential with increasein current density.

The current efficiencies under different electrolytic conditionsare measured by comparing the moles of carbon formed in theproduct to the Faradays of charge passed during the electrolysis(4F/mole of carbon). As shown in Fig. 5c, current efficiencies ofcarbon-production at different electrolytic conditions are in therange of ~71%e~82%. The highest current efficiency can reach over82% for the 577 �C LiNaK eutectic carbonate electrolyte at conditionof 200mA/cm2 electrolysis current density. As the increase in eithertemperature or current density would promote side reactions, suchas the reduction of alkali metal ions (Liþ, Naþ, Kþ), formation of CO,as well as other possible electrochemical reactions [35], the currentefficiency decreases notably with increasing electrolytic tempera-ture and current density. In spite of this, the current efficiency stillremains over 71% within the scope of the conditions investigated.

3.4. Li compared to LiNaK carbonate electrolyte: nanotubeimprovements in current efficiency, yield and voltage

The LiNaK eutectic (1) demonstrates the transition from a pallet/honeycomb carbon structured product to a nanomaterial productincluding nanotubes with increasing electrolytic temperature, and(2) provides a more cost effective electrolyte than pure lithiumcarbonate in which to conduct the electrolysis. However, the LiNaKeutectic carbonate presents two additional challenges not observedin the pure lithium carbonate (1) the small, but significant decrease

Table 1BET surface area (m2/g) of the carbon powders correlated with observed (by SEM) paLi2CO3eNa2CO3eK2CO3 at different electrolysis temperatures.

Temperature 577 �C 602 �C

Particle size (nm) 850e1200 1500e1800Surface area (m2/g) 198 31

in current efficiency observed in Fig. 5c for the transformation ofCO2 to carbon products, and (2) the small yield of the carbon thatcomprises the desired CNT product.

As we have recently reported [32], the conditions to improveCNT yield and uniformity in the pure lithium carbonate electrolyteinclude (1) the use of a Zn coated (galvanized) steel cathode as theZn lowers the initial carbon forming energy [51], (2) the controlledrelease of nickel from the anode, to serve as a cathode CNT nucle-ating agent, and (3) current density control in which an initial lowcurrent to allow nickel nucleation is followed by a higher currentdensity for controlled CNT growth [32]. In these experiments, theelectrolyte was contained in a 100 ml Ni crucible, and the innerwalls of the crucible also served as the (oxygen generating) anode.The cathode is formed by coiling 1.2 mm diameter galvanized (zinccoated) steel wire into a flat disc of 5 cm2 facing the lower floor ofthe crucible. The applied constant electrolysis current is graduallyincreased, 0.06A (for 12min), then 0.11A (for 9 min), then 0.31A (for5 min), followed by the extended high current (200mA/cm�2) of1A. The coulombic efficiency is over 80% (and approaches 100%with carefully recovery of all product after washing), the product(after washing off the electrolyte) consists of >80% pure carbonnanofibers [32]. As shown in the top left of Fig. 6, in Li2CO3 after twohours of electrolysis (0.4 Ah/cm�2) at 770 �C, straight carbonnanotubes are formed. The only procedural difference between thisand prior [32] syntheses is Ni, rather than galvanized steel wire isused as the cathode, and ZnO (1% by wt) is added directly to theelectrolyte, observed electrolysis potentials are low (1.5 V) at200 mA cm�2, and it is evident that as with the galvanized steelcathode, the Ni cathode forms high quality CNTs. The bright spotson the SEM in the figure are identified as Ni (nucleation sites) byEDS. As a new finding, an extended electrolysis duration (20 h,4.0 Ah/cm�2 on the galvanized steel cathode) generates widerdiameter carbon nanotubes as shown on the right top side of thefigure.

The LieNaeK carbonate is not observed to form CNTs in goodyield. Even under optimized (0.4 Ah/cm�2 on the galvanized steelcathode) conditions, electrolyses in the LieNaeK, rather than thepure Li, there is only a small yield (<15%) of CNTs as seen in thelower portion of Fig. 6, with the bulk of the product comprise ofgraphite platlets 150e250 nm thick and ~10 mm wide.

3.5. LiNa and Li Ba compared to LiNaK carbonate electrolyte

Under conditions delineated in the previous section, electrolysisin pure lithium carbonate provides a high yield of carbon nano-tubes. However, alternate carbonate electrolytes may offer advan-tages such as the greater natural abundance of sodium andpotassium compared to lithium carbonate, or the possibility fornovel carbon nanostructures. However, as also as delineated in theprevious section, the LiNaK carbonate electrolyte provides only amarginal yield of carbon nanotube products. Futhermore, weobserve that a 50/50 wt% mix of sodium and potassium carbonatesgenerates alkali metals rather than carbon at the cathode during a200 mA cm�2 electrolysis and that the product ignites a fire whenthe cathode is removed into humid air subsequent to the electrol-ysis. These challenges are overcome in Fig. 7 in which a LiNa

rticle size (nm) for the carbon product prepared at 400 mA/cm2 in 61:22:17 wt%

627 �C 652 �C 677 �C

2100e2450 2500e2700 3650e415020 15 11

Fig. 5. (a) and (b): Cell voltageetime plots of the electrolysis system at the indicated current densities. (c): Current efficiency of carbon-production under different electrolyticconditions. (A colour version of this figure can be viewed online.)

Fig. 6. SEM images of carbon materials obtained at current density of 200 mA/cm2 (subsequent to low current density nucleation) in various 770 �C electrolytes. Top: In purelithium carbonate for 2 h (left) or 20 h (right) electrolyses. Bottom: In LieNaeK CO3 carbonate for 2 h electrolyses.

H. Wu et al. / Carbon 106 (2016) 208e217214

carbonate electrolyte (rather than a NaK or LiNaK carbonate elec-trolyte) is investigated and provides both a more cost effectiveelectrolyte than the pure lithium carbonate, and importantly alsoproduces the high valued added carbon nanotube product in highyield, and at a high current efficiency and low energy (low elec-trolytic potential). The high value of the carbon nanotube productprovides a strong incentive to mitigate climate by removing thegreenhouse gas carbon dioxide and transforming it by electrolyticsplitting to carbon nanotubes in molten carbonate.

Fig. 7 includes CNT synthesized by electrolysis in 6.8 g of Na2CO3

mixedwith 70.0 g of Li2CO3 at 770 �C (8.9wt% Na2CO3) as presentedin SEMs in the top row, or in the middle row with 34.6 g of Na2CO3

mixedwith 34.1 g of Li2CO3 at 770 �C (~50wt% Na2CO3). Electrolysisin the 9 wt% Na2CO3 electrolyte generates a product comprising80e85% CNT. In this synthesis, and as with the high CNT yield pureLi2CO3 electrolyte syntheses, the electrolyte was contained in a100ml Ni crucible, and the inner walls of the crucible also served asthe (oxygen generating) anode. The cathode is a 5 cm2 galvanizedsteel coiled wire, and prior to the high, constant CNT growth cur-rent, an initial low current is applied to encourage CNT nucleation

sites. The applied constant electrolysis current was graduallyincreased, 0.06A (for 12 min), then 0.11A (for 9 min, at ~1V), then0.31A (for 5 min, at ~ 1.05V), followed by the extended high current(200 mA/cm2) of 1A for 90 min. During this time, the electrolyticpotential diminished from 2.2V to 2.0V in the initial 15 min, andthen gradually to a final electrolytic potential of 1.9V (which ishigher than observed in the pure Li2CO3 electrolyte in the previoussection), the decreasing potential presumably due to the increasingsurface area of the generated product which would tend towardslower polarization overpotential. This 1.9e2V potential at 200 mA/cm2 electrolysis current at 770 �C is considerably less than the 2.3to 2.5 observed at the same current density, but at lower temper-ature, in the LiNaK eutectic electrolyte in Fig. 5a.

Initially a higher weight fraction of Na2CO3 in a mixed electro-lyte with lithium carbonate, provided only a lower CNT productyield (~30%, not shown). However, as shown in the SEM in themiddle panels of Fig. 7, subsequent to 770 �C 200 mA cm�2 elec-trolysis in 50 wt% Na2CO3 at a galvanized steel cathode, it is evidentthat a high carbon nanotube product yield can be retained. Theproduct is more tangled than obtained from the pure Li2CO3

Fig. 7. SEM of high yield carbon nanotube products electrosynthesized at 770 �C from electrolytes other than pure Li2CO3 at lower (left) and higher (right) magnification. Top: in aLi2CO3 electrolyte containing 8.9 wt% Na2CO3. Middle in a Li2CO3 electrolyte containing nearly half (47.4 wt%) Na2CO3. Bottom in a Li2CO3 electrolyte containing 20 wt% BaCO3.

H. Wu et al. / Carbon 106 (2016) 208e217 215

electrolyte (and resembles the tangled product obtained in a Li2CO3with added Li2O [32]). The SEM exhibit the carbon product is over85% CNTs. The coulombic efficiency of the product is not as good(60%) as in the pure Li2CO3 electrolyte. Further optimization ofnucleation conditions is expected to increase both the yield andcoulombic efficiency.

Rather than the ~30% CNT yield, the >85% CNT product yield isachieved in the 50 wt% Na/Li composite carbonate electrolytewhenthe nickel concentration was constrained during the electrolysis bydecreasing the anode area (from the nickel crucible walls) to a10 cm2 Ni wire (and containing the electrolysis in an alumina

crucible). Solubility of nickel oxide is low in the electrolyte. Aspreviously noted nickel cation solubility increases as sodium andpotassium carbonate is added to the lithium carbonate electrolyte,and in particular a pure sodium/potassium carbonate electrolytecan be quite corrosive to the anode (the nickel anode dissolvesduring electrolysis and the electrolyte turns from transparent togreen) [57]. In lithium carbonate electrolytes, a small amount ofnickel oxide dissolves in the electrolyte as a stabilized electro-catalytic nickel oxide layer forms on the anode surface [18,58]. Thissmall amount of nickel cation migrates to the cathode where it isreduced forming nickel nucleation sites for CNF and CNT growth

H. Wu et al. / Carbon 106 (2016) 208e217216

[32,50] We have previously demonstrated that controlling theconcentration of Ni, by limiting Ni dissolution from the anode,controls and optimizes the number of transitional metal nucleationsites (in this case Ni re-deposited by reduction) on the cathode toenhance CNT yield [32]. Interestingly, the CNTs shown in themiddle of Fig. 7 are grownwithout the use or need of an initial lowcurrent step. This is consistent with a better nickel nucleationenvironment near the cathode due to a higher nickel cation solu-bility in the mixed Na/Li carbonate compared to the pure Li car-bonate electrolyte. This opens a pathway to use a less expensivethan pure Li2CO3 electrolyte for CO2 to CNT transformation.

A 20/80 wt% BaCO3/Li2CO3 electrolytewas also investigated, andit is evident in the lowest panel of Fig. 7 that a high yield carbonnanotube product is obtained at 770 �C and the same 200 mA cm�2

current density. We have previously observed that the solubility ofnickel is low in a mixed Li/Ba carbonate composite electrolyte [34].The synthesis was conducted under a hybrid of the electrolysisconditions in the pure or Li/Na carbonate syntheses (again usingthe initial low current density nucleation steps used in the pureLi2CO3 synthesis, but with the smaller coiled Ni wire cathode usedin the Na/LiCO3 synthesis). Interestingly, as shown in the figure, theCNTs are larger than obtained from pure Li2CO3, and also exhibit alarger range of CNT diameters than obtained from the pure Li2CO3

electrolyte. The SEM exhibit the carbon product is over 85% CNTs,although coulombic efficiency of the product is again not as good(55%) as in the pure lithium carbonate electrolyte. Further opti-mization of electrolyte fraction, nucleation conditions is expectedto increase both the yield and coulombic efficiency.

In the future we will report on the addition of oxides to theseand other composite carbonate electrolyte. We have previouslyobserved that oxide addition leads to further entanglement of theCNTs [32], and such defect-riddled tangled CNT structures cansignificantly improve Li-anode and Na-anode rechargeable batterycharge storage capabilities [50].

4. Conclusions

Here, we present a high yield pathway of generating carbonnanomaterials via molten salt electrolysis, driven by solar power aselectronic charge stored within Li-ion batteries. Honeycomb-likestructures are observed at 577 �C, which gradually transform intonanotubular structures with increasing electrolytic temperature.Carbon products with larger particle size and lower BET surfacearea are prepared at elevated electrolytic temperature, whereasincreased current densities favor the generation of carbon productsof lower particle size and higher BET surface area. Moreover, thecurrent efficiency of carbon production is increased as temperatureand current density decrease, reaching up to 82% under electrolyticcondition of 577 �C and 200 mA/cm2. Alkali metals, rather thancarbon products are formed in a mixed Na2CO3/K2CO3 electrolyte(with lithium salts). While the yield of carbon is high, the yield ofcarbon nanotubes as the specific carbon product formed at thecathode is marginal in a LiNaK carbonate eutectic. However, theyield of carbon nanotubes as the specific carbon product formed atthe cathode is high in each of (i) Li2CO3, (ii) mixed Li2CO3/Na2CO3and (iii) mixed Li2CO3/BaCO3 electrolytes; that is the CNT productyield is high in carbonate electrolytes containing lithium salts, butnot containing high concentrations of potassium salts. Ni, as well asgalvanized steel, is shown to be effective as a cathode to form CNTs,and long term electrolyses are shown to generate larger diameterCNTs.

Acknowledgements

This work is supported by the National Natural Science

Foundation of China (No. 21476046 and 21306022), SpecializedResearch Fund for the Doctoral Program of Higher Education ofChina (No. 20132322120002), China Postdoctoral Science Founda-tion (No. 2013M540269) and Postdoctoral Science Foundation ofHeilongjiang Province of China (No. LBH-TZ0417). SL is grateful forsupport of this research by the US National Science Foundationgrant 1505830. 770 �C LieNaeK, Li (both short and long-term),LieK, LieNa, LieBa carbonate experiments were conceived andperformed at GWU. Lower temperature LieNaeK carbonate ex-periments were conceived and performed at NPU & ECNU.

References

[1] D. Chery, V. Albin, V. Lair, M. Cassir, Thermodynamic and experimentalapproach of electrochemical reduction of CO2 in molten carbonates, Int. J.Hydrogen Energy 39 (23) (2014) 12330e12339.

[2] S. Licht, B. Wang, S. Ghosh, H. Ayub, D. Jiang, J. Ganley, A new solar carboncapture process: solar thermal electrochemical photo (STEP) carbon capture,J. Phys. Chem. Lett. 1 (15) (2010) 2363e2368.

[3] W.H.A. Peele, K. Hemmes, J.H.W. DeWit, CO2 reduction in molten 62/38 mole%Li/K carbonate mixture, Electrochim Acta 43 (7) (1998) 763e769.

[4] W. Lin, H. Frei, Photochemical CO2 splitting by metal-to-metal charge-transferexcitation in mesoporous ZrCu (I)-MCM-41 silicate sieve, J. Am. Chem. Soc.127 (6) (2005) 1610e1611.

[5] M. Cassir, S.J. McPhail, A. Moreno, Strategies and new developments in thefield of molten carbonates and high-temperature fuel cells in the carbon cycle,Int. J. Hydrogen Energy 37 (24) (2012) 19345e19350.

[6] P. Claes, D. Moyaux, D. Peeters, Solubility and solvation of carbon dioxide inthe molten Li2CO3/Na2CO3/K2CO3(43.5: 31.5: 25.0 mol%) eutectic mixture at973 K I. Experimental part, Eur. J. Inorg. Chem. 1999 (4) (1999) 583e588.

[7] S.C. Roy, O.K. Varghese, M. Paulose, C.A. Grimes, Toward solar fuels: photo-catalytic conversion of carbon dioxide to hydrocarbons, Acs Nano 4 (3) (2010)1259e1278.

[8] L. Kleiminger, G.H. Kelsall, T. Li, K. Li, Effects of current collector materials onperformances of micro-tubular solid oxide electrolysers for splitting CO2, ECST68 (1) (2015) 3449e3458.

[9] L. Gu, Y. Zhang, Unexpected CO2 splitting reactions to form CO with N-het-erocyclic carbenes as organocatalysts and aromatic aldehydes as oxygen ac-ceptors, J. Am. Chem. Soc. 132 (3) (2009) 914e915.

[10] D.S. Laitar, P. Müller, J.P. Sadighi, Efficient homogeneous catalysis in thereduction of CO2 to CO, J. Am. Chem. Soc. 127 (49) (2005) 17196e17197.

[11] J. Hawecker, J.M. Lehn, R. Ziessel, Photochemical and electrochemical reduc-tion of carbon dioxide to carbon monoxide mediated by (2, 20Bipyridine)tricarbonylchlororhenium (I) and related complexes as homogeneous cata-lysts, Helv. Chim. Acta 69 (8) (1986) 1990e2012.

[12] C. Bruckmeier, M.W. Lehenmeier, R. Reithmeier, B. Rieger, J. Herranz,C. Kavakli, Binuclear rhenium (I) complexes for the photocatalytic reduction ofCO2, Dalton Trans. 41 (16) (2012) 5026e5037.

[13] J. Chauvin, F. Lafolet, S. Chardon-Noblat, A. Deronzier, M. Jakonen, M. Haukka,Towards New molecular photocatalysts for CO2 reduction: photo-inducedelectron transfer versus CO dissociation within [Os (NN)(CO)2Cl2] com-plexes, Chem-A Eur. J. 17 (15) (2011) 4313e4322.

[14] S. Sato, T. Morikawa, T. Kajino, O. Ishitani, A highly efficient mononucleariridium complex photocatalyst for CO2 reduction under visible light, Angew.Chem. 125 (3) (2013) 1022e1026.

[15] K. Adachi, K. Ohta, T. Mizuno, Photocatalytic reduction of carbon dioxide tohydrocarbon using copper-loaded titanium dioxide, Sol. Energy 53 (2) (1994)187e190.

[16] H. Fei, M.D. Sampson, Y. Lee, C.P. Kubiak, S.M. Cohen, Photocatalytic CO2reduction to formate using a Mn (I) molecular catalyst in a robust metal-eorganic framework, Inorg. Chem. 54 (14) (2015) 6821e6828.

[17] J. Agarwal, E. Fujita, H.F. Schaefer III, J.T. Muckerman, Mechanisms for COproduction from CO2 using reduced rhenium tricarbonyl catalysts, J. Am.Chem. Soc. 134 (11) (2012) 5180e5186.

[18] Y. Kou, Y. Nabetani, D. Masui, T. Schimada, S. Takagi, H. Tachinaba, et al., Directdetection of key reaction intermediates in photochemical CO2 reductionsensitized by a rhenium bipyridine complex, J. Am. Chem. Soc. 136 (16) (2014)6021e6030.

[19] B. Gholamkhass, H. Mametsuka, K. Koike, T. Tanabe, M. Furue, O. Ishitani,Architecture of supramolecular metal complexes for photocatalytic CO2reduction: ruthenium-rhenium bi-and tetranuclear complexes, Inorg. Chem.44 (7) (2005) 2326e2336.

[20] T. Morimoto, C. Nishiura, M. Tanaka, J. Rohacova, Y. Nakagawa, Y. Funada, etal., Ring-shaped Re (I) multinuclear complexes with unique photofunctionalproperties, J. Am. Chem. Soc. 135 (36) (2013) 13266e13269.

[21] Y. Tamaki, T. Morimoto, K. Koike, O. Ishitani, Photocatalytic CO2 reductionwith high turnover frequency and selectivity of formic acid formation usingRu (II) multinuclear complexes, Proc. Natl. Acad. Sci. U. S. A. 109 (39) (2012)15673e15678.

[22] Y. Kitagawa, H. Takeda, K. Ohashi, T. Asatani, D. Kosumi, H. Hashimoto, et al.,Photochemical reduction of CO2 with Red light using synthetic

H. Wu et al. / Carbon 106 (2016) 208e217 217

chlorophyllerhenium bipyridine dyad, Chem. Lett. 43 (8) (2014) 1383e1385.[23] K. Maeda, R. Kuriki, M. Zhang, X. Wang, O. Ishitani, The effect of the pore-wall

structure of carbon nitride on photocatalytic CO2 reduction under visible light,J. Mater Chem. A 2 (36) (2014) 15146e15151.

[24] H. Takeda, H. Koizumi, K. Okamoto, O. Ishitani, Photocatalytic CO2 reductionusing a Mn complex as a catalyst, Chem. Commun. 50 (12) (2014) 1491e1493.

[25] A. Nakada, K. Koike, T. Nakashima, T. Morimoto, O. Ishitani, Photocatalytic CO2reduction to formic acid using a Ru (II)eRe (I) supramolecular complex in anaqueous solution, Inorg. Chem. 54 (4) (2015) 1800e1807.

[26] S. Neatu, J.A. Maci�a-Agull�o, P. Concepci�on, H. Garcia, Goldecopper nanoalloyssupported on TiO2 as photocatalysts for CO2 reduction by water, J. Am. Chem.Soc. 136 (45) (2014) 15969e15976.

[27] J. Gu, A. Wuttig, J.W. Krizan, Y. Hu, Z.M. Detweiler, R.J. Cava, et al., Mg-dopedCuFeO2 photocathodes for photoelectrochemical reduction of carbon dioxide,J. Phys. Chem. C 117 (24) (2013) 12415e12422.

[28] Q. Shen, Z. Chen, X. Huang, M. Liu, G. Zhao, High-yield and selective photo-electrocatalytic reduction of CO2 to formate by metallic copper decoratedCo3O4 nanotube arrays, Environ. Sci. Technol. 49 (9) (2015) 5828e5835.

[29] Q. Kang, T. Wang, P. Li, L. Liu, K. Chang, M. Liu, et al., Photocatalytic reductionof carbon dioxide by hydrous hydrazine over AueCu alloy nanoparticlessupported on SrTiO3/TiO2 coaxial nanotube arrays, Angew. Chem. 127 (3)(2015) 855e859.

[30] A.L. Dipu, Y. Ujisawa, J. Ryu, Y. Kato, Electrolysis of carbon dioxide for carbonmonoxide production in a tubular solid oxide electrolysis cell, Ann. Nucl.Energy 81 (2015) 257e262.

[31] V. Kaplan, E. Wachtel, K. Gartsman, Y. Feldman, I. Lubomirsky, Conversion ofCO2 to CO by electrolysis of molten lithium carbonate, J. Electrochem Soc. 157(4) (2010) B552eB556.

[32] J. Ren, F.F. Li, J. Lau, L. Gonz�alez-Urbina, S. Licht, One-pot synthesis of carbonnanofibers from CO2, Nano Lett. 15 (9) (2015) 6142e6148.

[33] K. Le Van, H. Groult, F. Lantelme, M. Dubois, D. Avignant, A. Tressaud, et al.,Electrochemical formation of carbon-nanopowders with various porosities inmolten alkali carbonates, Electrochim Acta 54 (19) (2009) 4566e4573.

[34] F.-F. Li, S. Liu, B. Cui, J. Lau, G. Stuart, S. Licht, A one-pot synthesis of hydrogenand carbon fuels from water and carbon dioxide, Adv. Energy Mat. 7 (2015),1401791-1-7.

[35] J. Ren, J. Lau, J. Lau, M. Lefler, S. Licht, The minimum electrolytic energy neededto convert carbon dioxide by electrolysis in carbonate melts, J. Phys. Chem. C119 (2015) 23342e23349.

[36] H.V. Ijije, R.C. Lawrence, G.Z. Chen, Carbon electrodeposition in molten salts:electrode reactions and applications, RSC Adv. 4 (67) (2014) 35808e35817.

[37] H.V. Ijije, R.C. Lawrence, N.J. Siambun, S.M. Jeong, D.A. Jewell, D. Hu, G.Z. Chen,Electro-deposition and re-oxidation of carbon in carbonate containing moltensalts (Open Access Article), Faraday Dis. 172 (2014) 105e116.

[38] H.V. Ijije, C.- Sun, G.Z. Chen, Indirect electrochemical reduction of carbon di-oxide to carbon nanopowders in molten alkali carbonates: process variablesand product properties, Carbon 73 (2014) 163e174.

[39] D. Tang, H. Yin, X. Mao, W. Xiao, D. Wang, Effects of applied voltage andtemperature on the electrochemical production of carbon powders from CO2

in molten salt with an inert anode, Electrochim Acta 114 (2013) 567e573.[40] S. Licht, Efficient solar generation of hydrogen fuel e a fundamental analysis,

Electrochem. Comm. 4/10 (2002) 789e794.[41] S. Licht, Solar water splitting to generate hydrogen fuel: photothermal elec-

trochemical analysis, J. Phys. Chem. B 107 (2003) 4253e4260.[42] S. Licht, L. Halperin, M. Kalina, M. Zidman, N. Halperin, Electrochemical po-

tential tuned solar water splitting, Chem. Comm. 2003 (2003) 3006e3007.[43] S. Licht, STEP (Solar Thermal Electrochemical Photo) generation of energetic

molecules: a solar chemical process to end anthropogenic global warming,J. Phys. Chem. C 113 (36) (2009) 16283e16292.

[44] S. Licht, Efficient solar-driven synthesis, carbon capture, and desalinization,STEP: solar thermal electrochemical production of fuels, metals, bleach, Adv.Mater. 23 (47) (2011) 5592e5612.

[45] S. Licht, H. Wu, C. Hettige, B. Wang, J. Asercion, J. Lau, et al., STEP cement: solarthermal electrochemical production of CaO without CO2 emission, Chem.Commun. 48 (48) (2012) 6019e6021.

[46] B. Cui, S. Licht, Critical STEP advances for sustainable iron production, GreenChem. 15 (4) (2013) 881e884.

[47] S. Licht, B. Cui, B. Wang, STEP carbon capture-the barium advantage, J. CO2Util. 2 (2013) 58e63.

[48] S. Licht, Cui, B. Wang, F.F. Li, J. Lau, S. Liu, Ammonia synthesis by N2 and steamelectrolysis in molten hydroxide suspensions of nanoscale Fe2O3, Science 345(2014) 637e640.

[49] F.-F. Li, J. Lau, S. Licht, Sungas instead of syngas: efficient co-production of COand H2 from a single beam of sunlight, Adv. Sci. 2 (1500260) (2015) 1e5.

[50] S. Licht, A. Douglas, J. Ren, R. Carter, M. Lefler, C.L. Pint, Carbon nanotubesproduced from ambient carbon dioxide for environmentally sustainablelithium-ion and sodium-ion battery anodes, ACS Central Sci. 2 (2016)162e168.

[51] G. Dey, J. Ren, T. El-Ghazawhi, S. Licht, How does amalgamated Ni cathodeaffect carbon nanotube growth? A density functional theory study, RSC Adv.(2016) 4, http://dx.doi.org/10.1039/C6RA03460H.

[52] D. Lv, L. Xu, Z. Xu, Z. Ge, Application of thermogravimetry and differentialthermal analysis in catalytic CVD synthesis of carbon nanotubes e effect ofreaction temperature on the production of carbon nanotubes, Chin. J. Chem.Phys. 15 (2) (2002) 118e122.

[53] A.C. Ferrari, D.M. Basko, Raman spectroscopy as a versatile tool for studyingthe properties of graphene, Nat. Nanotechnol. 8 (4) (2013) 235e246.

[54] Service R, Tailpipe to tank, Science 349 (2015) 1158e1160.[55] cheaptubes.com, Industrial Grade Carbon Nanotubes. April 20, 2016 at: http://

www.cheaptubes.com/product-category/industrial-grade-carbon-nanotubes/.[56] B. Wang, D. Yuan, H. Wu, X. Sui, Theoretical analysis of electrodeposition

potentals of carbonates driven by solar energy, J. Daqing Pet. Inst. 34 (5)(2010) 92e96.

[57] Licht S. Stabilization of STEP electrolyses in lithium-free molten carbonates.arXiv 2012 (Sept. 16); 1209.3512: 1-4.

[58] S. Licht, H. Wu, STEP iron, a chemistry of iron formation without CO2 emis-sion: molten carbonate solubility and electrochemistry of iron ore impurities,J. Phys. Chem. C 115 (2011) 25138e25147.