Cs

SJa

ob

c

d

a

ARR1A

KHLMNS

1

crbppeLtlmLc

Rgf

0h

J. of Supercritical Fluids 73 (2013) 70– 79

Contents lists available at SciVerse ScienceDirect

The Journal of Supercritical Fluids

jou rn al h om epa ge: www.elsev ier .com/ locate /supf lu

ontinuous synthesis of lithium iron phosphate (LiFePO4) nanoparticles inupercritical water: Effect of mixing tee

eung-Ah Honga, Su Jin Kimb, Kyung Yoon Chungb, Myung-Suk Chunc, Byung Gwon Leea,aehoon Kima,d,∗

Supercritical Fluid Research Laboratory, Clean Energy Research Center, Korea Institute of Science and Technology (KIST), Hwarangno 14-gil 5, Seongbuk-gu, Seoul 136-791, Republicf KoreaCenter for Energy Convergence, KIST, Republic of KoreaSensor System Research Center, KIST, Republic of KoreaGreen School, Korea University, 5-1 Anam Dong, Seongbuk-gu, Seoul 136-701, Republic of Korea

r t i c l e i n f o

rticle history:eceived 27 September 2012eceived in revised form3 November 2012ccepted 13 November 2012

eywords:ydrothermal synthesis

a b s t r a c t

Continuous supercritical hydrothermal synthesis of olivine (LiFePO4) nanoparticles was carried out usingmixing tees of three different geometries; a 90◦ tee (a conventional Swagelok® T-union), a 50◦ tee, anda swirling tee. The effects of mixing tee geometry and flow rates on the properties of the synthesizedLiFePO4, including particle size, surface area, crystalline structure, morphology, and electrochemical per-formance, were examined. It was found that, when the flow rate increased, the particle size decreased;however, the discharge capacity of the particles synthesized at the high flow rate was lower due to theenhanced formation of Fe3+ impurities. The use of a swirling tee led to smaller-sized LiFePO4 particles

ithium iron phosphateixing tee geometryanoparticlesupercritical water

with fewer impurities. As a result, a higher discharge capacity was observed with particles synthesizedwith a swirling tee when compared with discharge capacities of those synthesized using the 90◦ and50◦ tees. After carbon coating, the order of initial discharge capacity of LiFePO4 at a current density of17 mA/g (0.1C) and at 25 ◦C was swirling tee (149 mAh/g) > 50◦ tee (141 mAh/g) > 90◦ tee (135 mAh/g).The carbon-coated LiFePO4 synthesized using the swirling tee delivered 85 mAh/g at 20C-rate andat 55 ◦C.

. Introduction

In response to decreasing petroleum reserves and growingoncern about global climate change, there is great enthusiasmegarding the development of rechargeable lithium secondaryatteries (LIBs) for powering more sustainable methods of trans-ortation and renewable energy storage devices, for example,lug-in hybrid electric vehicles, electric vehicles, and electricnergy storage systems. For such applications, it is essential forIBs to offer an excellent safety profile, high energy/power densi-ies, excellent cyclability, and low cost [1,2]. The ordered olivineithium iron phosphate (LiFePO4) has been considered as one of

ost promising cathode materials for large-scale application inIBs [3–5]. Its advantageous features include a high theoreticalapacity of 170 mAh/g, a flat voltage profile at ∼3.4 V versus Li+/Li,

∗ Corresponding author at: Supercritical Fluid Research Laboratory, Clean Energyesearch Center, Korea Institute of Science and Technology (KIST), Hwarangno 14-il 5, Seongbuk-gu, Seoul 136-791, Republic of Korea. Tel.: +82 2 958 5874;ax: +82 2 958 5205.

E-mail address: jaehoonkim@kist.re.kr (J. Kim).

896-8446/$ – see front matter. Crown Copyright © 2012 Published by Elsevier B.V. All rittp://dx.doi.org/10.1016/j.supflu.2012.11.008

Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved.

the low cost of the starting materials, environmental benignity,high tolerance to overcharge, and high thermal stability [6–9].However, there are some drawbacks of LiFePO4 that need to beaddressed in order to improve its potential for use in practical appli-cations. These include low electronic conductivity (∼10−10 S/cm)and sluggish Li+ ion diffusivity (10−14 to 10−17 cm2/S) through theolivine structure [10,11]; as a result, a significant decrease in capac-ity at higher discharge rates is often observed, making LiFePO4unsuitable for use in high-power battery applications. Recently,considerable efforts have been directed toward improving the rateperformance of LiFePO4: reduction of particle size or control ofparticle porosity to shorten the transport path length of Li+ [12,13],a conductive layer coating (e.g., carbon, conducting polymer,metal oxides) [14–17], or doping with cations/anions [9,18] toenhance the intrinsic electron and/or Li+ conductivity. Anotherobstacle to the commercial use of LiFePO4 is the need to developreliable and economically viable large-scale production methods.Although various methods have been proposed to synthesis

LiFePO4 that includes solid-state, sol–gel, co-precipitation, hydro-thermal/solvothermal, molten state, spray solution, microwave,emulsion drying and so forth, only a few of them are currentlyapplied in the commercial production of LiFePO4 [5].

ghts reserved.

ercriti

aathhpntriSmtahrLIucy

anatmsvmtifTaigefeoicctpdth

Tecc[totomasicd

S.-A. Hong et al. / J. of Sup

Supercritical hydrothermal synthesis (SHS) is a very promisinglternative to the conventional techniques for producing cathodend anode active materials [19]. The unique physical proper-ies of supercritical water, including extremely low viscosity,igh reactant diffusivity, zero surface tension, high reactivity, andigh supersaturation ratio of reaction intermediates, make it aromising medium for the production of highly crystalline andanosized particles [20–22]. In addition, the supercritical hydro-hermal method is environmentally friendly, fast, simple, andeadily scalable by the employment of continuous operation. Var-ous types of metal oxide nanoparticles have been produced byHS, including CeO2, CuO, TiO2, Fe2O3, NiO, ZrO2, and ZnO; theorphology and size distribution of fine particles can be con-

rolled by adjusting pH, metal salt concentration, temperature,nd pressure [22,23]. In recent years, the advantages of SHSave led to it being widely utilized to synthesize active mate-ials for use in LIBs. This includes LiCoO2 [24,25], LiMn2O4 [26],iNi1/3Co1/3Mn1/3O2 [27], LiFePO4 [28–33], and Li4Ti5O12 [34–36].n 2010, the first commercial plant for the production of LiFePO4sing continuous SHS was constructed in Korea [17]. This plant canontinuously produce LiFePO4 and has a capacity of 1000 tons perear.

In a typical continuous SHS, the design of the mixing tee plays critical role in determining the properties of the synthesizedanoparticles (e.g., particle size, particle size distribution, particlegglomeration) and in allowing continuous operation without par-icle deposition on the mixing zone [37–42]. A stream of aqueous

etal salt solution and a stream of supercritical water, which haveignificantly different fluid and flow properties (e.g., temperature,iscosity, density, flow velocity, flow Reynolds number) meet at theixing tee. As the temperature of the aqueous solution approaches

hat of near-critical water, the OH− concentration increases; thisn turn accelerates the formation of nanoparticle by hydrolysis,ollowed by dehydration steps. When a conventional Swagelok®

-union is used as the fluid mixer, asymmetric flow convergencend frequent particle agglomeration are often observed due to annsufficient mixing rate and heterogeneous nucleation followed byrowth at the wall of the mixer [37,42]. Therefore, considerablefforts are still being made to develop a mixing tee with the capacityor more reliable, more uniform mixing and less particle agglom-ration, which will result in the production of high-quality metalxide nanoparticles various nanoparticle applications. To date, var-ous types of mixing tees have been developed, including the centralollision-type micromixer [37], the T-type micromixer [38], theross-type mixer [39], the swirling micromixer [40], and the nozzle-ype mixer [42]. The use of these mixing tees has resulted in theroduction of smaller-sized particles with more uniform particleistribution, when compared with those produced by the conven-ional T-union mixer, owing to the improved mixing and more rapideat transfer.

In our previous studies, in which we utilized a Swagelok®

-union (90◦ tee) and a home-made 50◦ tee, we analyzed theffects of various process parameters (temperature, flow rates,oncentration) on the properties of LiFePO4 and the effects ofarbon coating on the electrochemical performance of LiFePO432,33]. The carbon-coated LiFePO4 (C-LiFePO4) synthesized usinghe aforementioned tees exhibited a rather low discharge capacityf ∼135 mAh/g. In the present study, a new mixing tee geome-ry, a swirling-type mixing tee, was evaluated for the productionf LiFePO4 nanoparticles with enhanced electrochemical perfor-ance. The following sections describe the effects of flow rate

nd mixing tee geometry on particle properties including particle

ize, surface area, morphology, crystallinity, and electrochem-cal performance. The explanation for the improved dischargeapacity of the C-LiFePO4 synthesized using the swirling mixer isiscussed.

cal Fluids 73 (2013) 70– 79 71

2. Experimental

2.1. Materials

Lithium hydroxide monohydrate (LiOH·H2O, purity of >98 wt%),iron sulfate heptahydrate (FeSO4·7H2O, purity of > 99 wt%), phos-phoric acid (H3PO4, purity of >98 wt%), and sucrose (C12H22O11,purity of ≥99 wt%) were purchased from Sigma–Aldrich (St. Louis,MO, USA) and used as received. Nitrogen (purity of >99.9%) andargon with 5% hydrogen (purity of >99.999%) were obtained fromShinyang Sanso Co. (Seoul, Korea). Distilled and deionized (DDI)water was prepared using a Milli-Q® Ultrapure water-purificationsystem with a 0.22 �m filter (Billerica, MA, USA). The cellulose estermembrane filter with a pore size of 0.45 �m was purchased fromToyo Roshi Kaisha Ltd. (Tokyo, Japan). Polyvinylidene difluoride(PVDF; Kureha Chem. Co., Tokyo, Japan), acetylene black (DENKACo. Ltd., Tokyo, Japan) and 1-methyl-2-pyrrolidinone (NMP; purityof ≥98 wt%, Alfa-Aesar, MA, USA) were used as received.

2.2. Continuous SHS apparatus and process

The continuous tubular high-pressure and high-temperatureapparatus was built in the Supercritical Fluid Research Labora-tory of the Korea Institute of Science and Technology for carryingout research into metal and metal oxide nanoparticle synthesis insupercritical water or in supercritical alcohols [43–46]. The geome-tries of the mixing tees tested in this work are presented in Fig. 1.The 90◦ tee is the commercially available Swagelok® T-union. In the50◦ tee, the flow of precursor solutions at room temperature meetsthe flow of supercritical water at a cross angle of 50◦. In the swirling-type mixing tee, the supercritical water flows are introduced fromtwo different directions, and each supercritical water flow mixeswith the precursor solution flow at a cross angle of 60◦. Shiftingthe supercritical water flow lines to a distance of 1.8 cm from theprecursor-solution flow line generates a swirling flow in the mixingtee.

The connector between the mixing tee and the reactor has thesame inner diameter of 6.5 mm and the whole volume of mix-ing fluids was introduced into the reactor. The inner volumes ofthe mixing tees were 1.5 cm3 (90◦ tee), 2.25 cm3 (50◦ tee), and4.5 cm3 (swirling tee). The flow pattern at the mixing tee wascalculated by computational fluid dynamics modeling using theFLUENT program (version 6.2). Details of the simulation condi-tions are given in Table S1 and the temperature distribution inthe mixing tees are given in Fig. S1 (supplementary data). In thesynthesis under investigation, the mass flow rate of supercriticalwater is 3–6 times higher than that of the precursor solution. TheRe of the precursor solution flows was in the range 229–952 (lam-inar), and that of the supercritical water flows was in the range15,100–16,400 (turbulent), as listed in Table 1. The mixed flow ofthe combined precursor solution and supercritical water down-stream of the mixing tee was turbulent with a Re in the range140,000–269,000. Typically, the temperature of the reactor andmixing tee was maintained at 400 ± 5 ◦C, the pressure of the wholesystem was kept at 25 ± 0.1 MPa, and the flow fluctuation waswithin ± 0.2 g/min during the synthesis. The mole concentrationratio of LiOH/H3PO4/FeSO4 was maintained at 0.09:0.03:0.03 inorder to keep the pH of the solution at ∼8 and maintain neutralor slightly basic conditions [28]. The obtained particles were puri-fied by being dispersed in DDI water, sonicated, and decanted using

centrifugation at 3000 rpm for 30 min. The purification procedurewas carried out in triplicate and the purified particles were driedat 60 ◦C in a vacuum oven for 24 h to remove the moisture in theLiFePO4 particles. The synthesis conditions are listed in Table 1.

72 S.-A. Hong et al. / J. of Supercritical Fluids 73 (2013) 70– 79

Fig. 1. Computational fluid dynamics simulation results of the three mixing tees. (a) 90◦ tee, (b) 50◦ tee and (c) swirling tee.

Table 1Synthesis conditions for LiFePO4.

Sample code Tee design Flow rates (g/min) Residence Time (s) Reynolds number (Re)

LiOH FeSO4·H3PO4 H2O Precursor solution Supercritical water After the mixingtee (total fluids)

E90-H 90◦ tee 3 3 18 18 560 16,400 269,000E90-M 90◦ tee 3 3 9 31 590 15,100 167,000E90-L 90◦ tee 1.7 1.7 9 37 291 15,700 140,000E50-H 50◦ tee 3 3 18 18 952 16,400 269,000E50-M 50◦ tee 3 3 9 32 937 15,100 168,000E50-L 50◦ tee 1.7 1.7 9 38 345 15,600 140,000ES-H Swirling tee 3 3 18 18 352 16,300 269,000ES-M Swirling tee 3 3 9 32 384 15,100 168,000ES-L Swirling tee 1.7 1.7 9 38 229 15,400 140,000

ercriti

2

stmiptahfp

2

tTHTeddOupbTwJc(

2

siwJagec1iccKbr

3

sTo1nrtctp

S.-A. Hong et al. / J. of Sup

.3. Carbon coating

In order to layer a carbon coating on the LiFePO4 particles,ucrose (as a carbon source) was dissolved in 1.3 ml DDI watero prepare a 12 wt% solution. Four grams of LiFePO4 particles was

ixed into the sucrose solution and the mixture was dried at 80 ◦Cn a vacuum oven for 24 h to evaporate the water. After the driedowders had been ground and strained using a 5-�m size sieve,he powder was sintered at 600 ◦C with a flow of 5% hydrogen inrgon at 100 ml/min for 3 h (heating rate of 5 ◦C/min). During theeat treatment under the reducing condition, a carbon layer formed

rom the precursor that was coated on the surface of the LiFePO4articles.

.4. Characterization

The structure of the particles was characterized by X-ray diffrac-ion (XRD) using a D/Max-2500 V/PC X-ray diffractometer (Rigaku,okyo, Japan). The morphology of the particles was observed using aitachi S-4100 field emission scanning electron microscope (SEM;okyo, Japan) and a Tecnai-F20 G2 high-resolution transmissionlectron microscope (HR-TEM; FEI Co. Ltd., OR, USA). The carbonistribution of the C-LiFePO4 samples was observed using energy-ispersive X-ray spectroscopy (EDX; model FP6595/05, FEI Co. Ltd.,R, USA). A copper grid coated with a silicon monoxide film wassed to ensure that any carbon detected originated from the sam-les. The carbon contents of the C-LiFePO4 samples were measuredy elemental analysis (model TC-136, LECO Corporation, MI, USA).he Brunauer–Emmett–Teller (BET) surface area of the particlesas measured using a BELSORP mini II apparatus (BEL Inc., Osaka,

apan). Elemental analyses of Li, Fe, and P in the samples werearried out using inductively coupled plasma mass spectroscopyICP-MS; ELAN 6100 series, Perkin-Elmer, NY, USA).

.5. Electrochemical measurements

For the electrochemical test of the bare LiFePO4 and C-LiFePO4amples, the active material (85 wt%), acetylene black as a conduct-ng material (10 wt%), and PVDF as a binder (5 wt%) in NMP were

ell mixed using a homogenizer (Nihonseiki Kaisha Ltd., Tokyo,apan). The cathodes were incorporated into cells with a lithium foilnode and a Celgard 2500 microporous membrane separator (Cel-ard LLC, Charlotte, NC, USA). The electrolyte was 1 M LiPF6 in anthylene carbonate (EC)/dimethyl carbonate (DMC)/ethylmethylarbonate (EMC) solvent with an EC/DMC/EMC volume ratio of:1:1. The cells were assembled in a dry room. Electrochem-

cal characterization was performed in a standard 2032 coinell configuration using a commercial multichannel galvanostaticharge–discharge cycler (model WBCS 3000, WonATech Corp.,orea) at temperatures of −20, 25, and 55 ◦C. The cells were cycledetween 2.5 V and 4.3 V versus Li+/Li at a 0.1–30C-rate (which cor-esponds to current densities of 17–5100 mAh/g).

. Results and discussion

Fig. 2 shows the XRD patterns of the LiFePO4 particles synthe-ized using the three different mixing tees at the various flow rates.he main diffraction patterns of the LiFePO4 can be indexed torthorhombic LiFePO4 olivine-type phase (JCPDS PDF number 40-499). The profiles of the peaks in the patterns are quite sharp andarrow, indicating that the LiFePO4 particles prepared using SHSetained a highly crystalline structure without additional calcina-

ion. When the continuous hydrothermal synthesis of LiFePO4 wasarried out in the subcritical water condition (300 ◦C and 25 MPa),he particles were highly agglomolated and the LiFePO4 crystallinehase did not form. Some impurity peaks of relatively low intensity

cal Fluids 73 (2013) 70– 79 73

can be observed in each sample; the peaks at 2� values of 33◦, 41◦,and 54◦ can be assigned to the (1 0 4), (1 1 3), and (1 1 6) diffractionplanes of Fe2O3 (Fe3+), and the peaks at 2� values of 30◦ and 43◦

can be assigned to the (2 2 0) and (4 0 0) diffraction planes of Fe3O4(Fe2+/Fe3+) phase. This indicates that some portion of the Fe2+ pre-cursor (FeSO4) is oxidized to the Fe3+ species during the SHS. It isworth noting that the whole reactor, the DDI water, and the aque-ous precursor solution were purged with high-purity nitrogen for1 h prior to the synthesis, and the precursor solution and waterreservoir were continuously purged during the synthesis to mini-mize oxygen content. During the synthesis, the oxygen content inthe precursor solution and water reservoir was measured to be verysmall (∼0.13 ppm). Thus, the formation of the Fe3+ impurities maybe due to an inherent preference of the Fe2+ precursor for oxidizingin the supercritical water condition. In fact, the particles obtainedfrom the SHS using the iron(II) precursor (FeSO4·7H2O) were foundto be mixed phase of magnetite (Fe3O4) and hematite (Fe2O3), asshown in Fig. S2. The formation rate of iron oxide species in super-critical water is known to be very fast owing to the low intermediatesolubility and high supersaturation ratio [38,47]. This may cause theformation of the Fe3+ impurity phase during SHS of LiFePO4.

The peak intensity ratio of the (1 0 4) Fe2O3 phase or the (4 0 0)Fe3O4 phase to the (1 1 1) LiFePO4 phase can serve as an indica-tor of the amount of Fe3+ impurities in the sample. As listed inTable 2, a relatively lower amount of impurities was present in thesamples synthesized using the swirling tee when compared withthose synthesized using the 50◦ and 90◦ tees at the condition ofmedium-to-low flow rate (M and L samples). The better mixingbetween the precursor solution flow and supercritical water flowin the swirling tee can induce more rapid nucleation of the parti-cles due to the lower intermediate solubility. Thus, the precursorsexperienced similar stage of nucleation, which may lead to betterchance to form LiFePO4 with lower amount of impurities. In addi-tion, the use of the swirling tee resulted in particles with highercrystallinity; the (0 4 0)(5 1 2), (1 1 3)(2 0 3), and (6 2 0) peaks of theLiFePO4 phase are clearly better split for the samples prepared usingthe swirling tee when compared with those for the samples pre-pared using the 50◦ or 90◦ tees; this difference is more profoundat the high-flow-rate condition. When the flow rate increased, thepeak intensity ratios of (1 0 4) Fe2O3 phase to (1 1 1) LiFePO4 phaseand/or of (4 0 0) Fe3O4 phase to (1 1 1) LiFePO4 phase increased,except the 50◦ case. The Fe content in each sample, measured byICP-MS, also increased with an increment in the flow rate. Thissuggests that larger amounts of the Fe3+ impurities formed at thehigh-flow-rate condition. Similar trend was observed in the con-tinuous synthesis of CoFe2O4 in hot-temperature water. Amountof the impurity (Fe2O3) was increased from 3% to 6% when the res-idence time decreased from 19 to 11 s at 295 ◦C and 24 MPa [48].The increase in iron oxide phase at high flow rate condition maybe due to much lower solubility and much higher supersaturationratio of iron intermediates when compared to other species (e.g.,Co intermediates, PO4 intermediates). For example, the formationrate of iron oxide nanoparticles in supercritical water is extremelyfast (0.002 s) and the conversion was very high (97.9%) due to thelow intermediate solubility in scH2O [38]. Thus at the high flowrate condition, iron precursor may precipitate in forms of Fe2O3and Fe3O4 when the other intermediate species still remain in thefluid phase.

Fig. 3 shows SEM images of the LiFePO4 particles synthesizedusing the three different mixing tees at different flow rates. Atthe high flow rate, the use of the swirling tee resulted in muchsmaller sized particles (ES-H) when compared with the particles

synthesized using the 90◦ tee (E90-H) and the 50◦ tee (E50-H).This leads to a larger BET surface area for ES-H than for E50-H andE90-H, as shown in Table 2. The size distribution of the sampleswas rather broad (E90-H: 200–600 nm; E50-H: 300–700 nm; ES-H:

74 S.-A. Hong et al. / J. of Supercritical Fluids 73 (2013) 70– 79

F 4/H3PF

1nrr

ig. 2. X-ray diffraction patterns of the LiFePO4 particles synthesized at LiOH/(FeSOe3O4; �: Fe2O3).

00–400 nm), suggesting that each particle experienced a differentucleation and growth stage during its formation. At lower flowates, the samples synthesized using the three different mixing teesetained very similar particle size. Indeed, the BET surface areas of

O4)/H2O flow rates of: (a) 3:3:18 g/min, (b) 3:3:9 g/min and (c) 1.7:1.7:9 g/min (•:

the samples synthesized at the low flow rate (E90-L, E50-L, ES-L)were very similar, in the range 6.2–7.3 m2/g. Regardless of mixingtee geometry, the size of the particles synthesized at the higher flowrate is much smaller than that of those synthesized at the lower flow

S.-A. Hong et al. / J. of Supercritical Fluids 73 (2013) 70– 79 75

Table 2Elemental composition, BET surface area, conductivity, carbon content, and initial/30th discharge capacity of LiFePO4 and carbon-coated LiFePO4.

Sample code Mole ratioa BET surfacearea (m2/g)

Peak intensity ratio Carbon content inC-LiFePO4

b (wt%)Conductivity(S/cm)

Initial/30thdischarge capacityc

(mAh/g)

Li Fe P (1 0 4 )Fe2O3/(1 1 1)LiFePO4

(4 0 0)Fe3O4/(1 1 1)LiFePO4

E90-H 0.49 0.72 0.50 15.9 0.1698 0.0984 0 – 52/43E90-M 0.55 0.69 0.52 7.5 0.1379 0.0425 0 – 65/55E90-L 0.50 0.59 0.46 6.2 0.1247 0.0435 0 1.1 × 10−9 85/73CE90-L 0.52 0.60 0.50 38.5 – – 5.91 7.1 × 10−5 135/125E50-H 0.31 0.65 0.45 12.2 0.1159 0.0917 0 – 66/57E50-M 0.53 0.66 0.51 6.3 0.1253 0.0363 0 – 76/67E50-L 0.49 0.61 0.51 7.0 0.1204 0.0441 0 1.1 × 10−9 100/85CE50-L 0.52 0.60 0.50 38.5 – – 5.90 7.1 × 10−5 141/135ES-H 0.39 0.63 0.45 17.4 0.1530 0.0873 0 – 78/67ES-M 0.33 0.62 0.46 10.4 0.1029 0.0407 0 – 88/72ES-L 0.49 0.59 0.49 7.3 0.0689 0.0458 0 1.2 × 10−9 100/87CES-L 0.47 0.59 0.48 41.2 – – 6.26 6.5 × 10−5 149/143

rs[Ri

a Analyzed by ICP-MS.b Analyzed by EA.c Initial and 30th discharge capacities at 0.1C.

ate. Several research groups have observed a reduction in particleize at higher flow rates with continuous SHS, for example CeO2

23], �-AlO(OH) [49], NiO [41], and ZnO [50]. As listed in Table 1,e at the high-flow-rate condition (E90-H, E50-H, ES-H) is approx-

mately two times higher than Re at the low-flow-rate condition

Fig. 3. Scanning electron microscope images of the LiFePO4 particles

(E90-L, E50-L, ES-L). The reduction in particle size at the higher flowrate can be attributed to improved mixing of fluids and rapid heat

transfer in the mixing region at the nucleation stage [42]. Thus,reaction intermediates can be consumed at the nucleation stage,resulting in smaller size particles. At the low flow rate condition

synthesized using the three mixing tees at various flow rates.

76 S.-A. Hong et al. / J. of Supercritical Fluids 73 (2013) 70– 79

F ples

( ischar

(od

scs

ig. 4. The cycling performance and the charge–discharge curves of the LiFePO4 same) 3:3:9 g/min; and (c), (f) 1.7:1.7:9 g/min (closed symbol: charge; open symbol: d

long reaction time), particle size is not dependent on the typesf mixing tee, suggesting dissolution and recrystallization may beominant particle formation mechanism.

Fig. 4 shows the electrochemical properties of the bare LiFePO4

amples measured at 25 ◦C. Table 2 lists the initial and 30th dis-harge capacities of each sample. All of the bare LiFePO4 samplesynthesized using the three different mixing tees show sloping

synthesized at the LiOH/(FeSO4/H3PO4)/H2O flow rates of (a), (d) 3:3:18 g/min; (b),ge).

voltage plateaus and low discharge capacities of equal to or lessthan 100 mAh/g at 0.1C-rate, indicating that the active materi-als are not properly utilized without carbon coating. In general,higher discharge capacities are observed when the LiFePO4 par-

ticles have higher crystallinity, smaller size, and a lower amountof impurities [51]. Under the higher-flow-rate condition (Fig. 4aand b), the discharge capacities were in the order swirling tee > 50◦

S.-A. Hong et al. / J. of Supercritical Fluids 73 (2013) 70– 79 77

F rsive( flow r

tscoctclcfldpwLrEa(o

LuopmmcTfiTo

ig. 5. High-resolution transmission electron microscope images and energy-dispeb), (e) the 90◦ tee; and (c), (d), (e) the swirling tee at the LiOH/(FeSO4/H3PO4)/H2O

ee > 90◦ tee. Smaller-sized particles and higher crystallinity in theamples synthesized using the swirling tee may result in higher dis-harge capacities because each sample retained a similar amountf impurities (Table 2). On the other hand, under the low-flow-rateondition (Fig. 4c), the order of discharge capacities was swirlingee ≥ 50◦ tee > 90◦ tee. ES-L and E50-L exhibited similar dischargeapacity patterns until the 15th cycle while ES-L showed slightlyarger capacities than E50-L after the 16th cycle. Since the parti-le sizes and crystallinity of each sample synthesized at the lowow rate were very similar, the impurities may play a role in theifference in discharge capacity. In fact, as shown in Table 2, theeak intensity ratio of (1 0 4) Fe2O3 phase to (1 1 1) LiFePO4 phaseas relatively lower for ES-L when compared with those of E90-

and E50-L. The lower amount of impurities in ES-L may also beesponsible for its smaller polarization; at a capacity of 40 mAh/g,S-L showed a much smaller voltage difference between chargend discharge curves (0.09 V) when compared with that of E50-L0.2 V), even though the initial capacity of E50-L was similar to thatf ES-L.

To improve the electrochemical performance of the bareiFePO4 samples, the particles were coated with a carbon layersing sucrose as the carbon source. As discussed in our previ-us work, carbon content, carbon thickness, and carbon structurelay an important role in determining the electrochemical perfor-ance of C-LiFePO4 synthesized by the supercritical hydrothermalethod as well as the solid-state method [33]. The optimum carbon

ontent for enhanced discharge capacity was found to be ∼6 wt%.

herefore, in this work, the carbon content of the C-LiFePO4 wasxed at ∼6 wt% by adjusting the sucrose concentration. As shown inable 2, the 6 wt% carbon coating on the LiFePO4 led to a four ordersf magnitude increase in conductivity from ∼10−9 to ∼10−5 S/cm.

X-ray spectroscopy analysis results of the LiFePO4 particles synthesized using (a),ate of 1.7:1.7:9 g/min.

Fig. 5 shows HR-TEM images, EDX results, and selected areaelectron diffraction patterns of the C-LiFePO4 samples synthesizedusing the 90◦ tee (CE90-L) and the swirling tee (CES-L). In theHR-TEM images, the LiFePO4 particles appear as dark regions andcarbon as light gray regions, as confirmed by the EDX analysis.Both of the C-LiFePO4 samples revealed individual LiFePO4 particlesembedded into the carbon network. Observation of the interfacebetween the LiFePO4 particle and the carbon layer showed that thecarbon distribution around the particle is not very uniform [33].The XRD patterns of the C-LiFePO4 samples shown in Fig. S4 revealthat the Fe3+ impurities in each sample disappeared after the car-bon coating. This may be because the Fe3+ impurities changed toiron phosphides (FeP, Fe2P, Fe3P) and/or iron carbide species (Fe3C,Fe2C) during the carbothermal reduction [33,52,53].

Fig. 6 shows the charge–discharge curves and cycling perform-ances of the C-LiFePO4 samples at 0.1 C-rate in the potential range2.5–4.3 V at 25 ◦C. The discharge capacities of the bare LiFePO4 sam-ples are shown in the figure for comparison purposes; the dischargecapacities of the C-LiFePO4 samples are significantly improvedwhen compared with those of the bare LiFePO4 particles, indi-cating that the carbon-coated active materials were utilized moreeffectively than the uncoated samples owing to their enhancedelectronic conductivity. After carbon coating, CES-L exhibited anextremely flat voltage curve shown at ∼3.4 and 3.5 V during dis-charge and charge while CE50-L and CE90-L exhibited slopingvoltage profiles. In addition, the discharge capacity of the C-LiFePO4prepared using the swirling tee was higher than those of the sam-

ples prepared using the 50◦ and the 90◦ tees; the initial dischargecapacities of the CES-L, CE50-L, and CE90-L samples were 135, 141,and 149 mAh/g, respectively, and their discharge capacities afterthe 30th cycle were 117, 134, and 143 mAh/g, respectively. Again,

78 S.-A. Hong et al. / J. of Supercriti

Fbo

tts

rtT

F(

ig. 6. (a) Charge–discharge voltage profiles and (b) cycling performance of theare LiFePO4 and C-LiFePO4 synthesized at the LiOH/(FeSO4/H3PO4)/H2O flow ratef 1.7:1.7:9 g/min (closed symbol: charge; open symbol: discharge).

he lower amount of impurities in the samples synthesized usinghe swirling tee under the low-flow-rate condition may be respon-ible for the higher discharge capacity.

To investigate potential high-power, outdoor applications, theate capabilities of CES-L were measured at various current densi-ies and different temperatures, and the results are shown in Fig. 7.he sample was progressively charged and discharged at various

ig. 7. Rate performance of CES-L at different C-rates and at different temperaturesclosed symbol: charge; open symbol: discharge).

cal Fluids 73 (2013) 70– 79

C-rates from 0.1 to 30C-rate for 55 cycles and then again at 0.1C-ratefor 10 cycles. The discharge capacity of CES-L was very stable at eachcurrent rate and at each temperature. At the high 20C-rate (whichmeans that it takes 3 min to charge and discharge), CES-L demon-strated a relatively high discharge capacity of 85 mAh/g at 55 ◦Cand 73 mAh/g at 25 ◦C, and a low discharge capacity of 20 mAh/g at−20 ◦C. The higher discharge capacity of LiFePO4 observed at hightemperature is due to the faster lithium diffusion rate in LiFePO4[54]. When returning to 0.1C-rate after 55 charge–discharge cycles,the discharge capacities were 150, 143, and 115 mAh/g at 55, 25,and −20 ◦C, respectively, indicating only ∼3% capacity loss fromthe initial discharge capacities. This indicates that the high crys-talline olivine phase of CES-L can retain its structural integrity evenduring the high Li+ ion intercalation and de-intercalation process.The results observed in this study may indicate that the use of theswirling tee in SHS could be a promising method for the productionof LiFePO4 as a cathode material for lithium ion batteries.

4. Conclusion

Mixing tees of three different geometries, 90◦, 50◦, and swirlingtees, were evaluated in continuous SHS for the production ofnanosized LiFePO4 particles with improved electrochemical prop-erties. Use of the swirling tee resulted in smaller-sized particleswith higher discharge capacities and lower polarization under thehigh-flow-rate condition. When the flow rate decreased, particlesize increased. Even though similarly sized particles in the range400–900 nm were produced using the three mixing tees at the lowflow rate, a smaller amount of impurities was present in the parti-cles produced by the swirling tee when compared with those fromthe 90◦ and 50◦ tees. As a result, a higher discharge capacity wasobserved with samples produced with the swirling tee. After car-bon coating, the discharge capacities of C-LiFePO4 at 0.1C-rate andafter 30 cycles, measured at 25 ◦C, were 143 mAh/g (swirling tee),135 mAh/g (50◦ tee), and 125 mAh/g (90◦ tee). When the dischargecapacity of the sample produced with the swirling tee was mea-sured at the higher temperature of 55 ◦C, the value was 155 mAh/g(which corresponds to 92% of theoretical value) at 0.1C-rate and103 mAh/g at 10C-rate. These results, showing highly crystallineLiFePO4 particles with smaller amounts of impurities and betterdischarge capacities, suggest that the swirling tee may be a verypromising alternative to conventional tees in the SHS of LiFePO4.

Acknowledgments

This research was supported by the KIST Young Fellow Programof the Korea Institute of Science and Technology. The authors alsoacknowledge support from the Global Research Lab (GRL) Programthrough the National Research Foundation of Korea (NRF) fundedby the Ministry of Education, Science and Technology (MEST) (grantnumber: 2011-00115) and the NRF of Korea Grant funded by theKorean Government (MEST) (2012, University-Institute coopera-tion program).

Appendix A. Supplementary data

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.supflu.2012.11.008.

References

[1] B. Scrosati, J. Hassoun, Y.K. Sun, Lithium-ion batteries. A look into the future,Energy & Environmental Science 4 (2011) 3287–3295.

[2] V. Etacheri, R. Marom, R. Elazari, G. Salitra, D. Aurbach, Challenges in the devel-opment of advanced Li-ion batteries: a review, Energy & Environmental Science4 (2011) 3243–3262.

ercriti

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

S.-A. Hong et al. / J. of Sup

[3] O.K. Park, Y. Cho, S. Lee, H.C. Yoo, H.K. Song, J. Cho, Who will drive electric vehi-cles, olivine or spinel? Energy & Environmental Science 4 (2011) 1621–1633.

[4] G. Jeong, Y.U. Kim, H. Kim, Y.J. Kim, H.J. Sohn, Prospective materials and appli-cations for Li secondary batteries, Energy & Environmental Science 4 (2011)1986–2002.

[5] J.J. Wang, X.L. Sun, Understanding and recent development of carbon coating onLiFePO4 cathode materials for lithium-ion batteries, Energy & EnvironmentalScience 5 (2012) 5163–5185.

[6] A.S. Arico, P. Bruce, B. Scrosati, J.M. Tarascon, W. Van Schalkwijk, Nanostruc-tured materials for advanced energy conversion and storage devices, NatureMaterials 4 (2005) 366–377.

[7] M. Armand, J.M. Tarascon, Building better batteries, Nature 451 (2008)652–657.

[8] H. Huang, S.-C. Yin, L.F. Nazar, Approaching theoretical capacity of LiFePO4

at room temperature at high rates, Electrochemical and Solid State Letters 4(2001) A170–A172.

[9] S.-Y. Chung, J.T. Bloking, Y.-M. Chiang, Electronically conductive phospho-olivines as lithium storage electrodes, Nature Materials 1 (2002) 123–128.

10] C. Wang, J. Hong, Ionic/electronic conducting characteristics of LiFePO4 cathodematerials, Electrochemical and Solid State Letters 10 (2007) A65–A69.

11] P.P. Prosini, M. Lisi, D. Zane, M. Pasquali, Determination of the chemical diffu-sion coefficient of lithium in LiFePO4, Solid State Ionics 148 (2002) 45–51.

12] A. Yamada, S.C. Chung, K. Hinokuma, Optimized LiFePO4 for lithium batterycathodes, J. The Electrochemical Society 148 (2001) A224–A229.

13] C. Delacourt, P. Poizot, S. Levasseur, C. Masquelier, Size effects on carbon-freeLiFePO4 powders: the key to superior energy density, Electrochemistry SolidState Letters 9 (2006) A352–A355.

14] Y.G. Wang, Y.R. Wang, E.J. Hosono, K.X. Wang, H.S. Zhou, The design of aLiFePO4/carbon nanocomposite with a core–shell structure and its synthesisby an in situ polymerization restriction method, Angewandte Chemie Interna-tional Edition 47 (2008) 7461–7465.

15] K.S. Park, S.B. Schougaard, J.B. Goodenough, Conducting-polymer/iron-redox-couple composite cathodes for lithium secondary batteries, Advanced Materials19 (2007) 848–851.

16] Y.S. Hu, Y.G. Guo, R. Dominko, M. Gaberscek, J. Jamnik, J. Maier, Improvedelectrode performance of porous LiFePO4 using RuO2 as an oxidic nanoscaleinterconnect, Advanced Materials 19 (2007) 1963–1966.

17] S. Yoon, C. Liao, X.G. Sun, C.A. Bridges, R.R. Unocic, J. Nanda, S. Dai, M.P.Paranthaman, Conductive surface modification of LiFePO4 with nitrogen-doped carbon layers for lithium-ion batteries, J. Materials Chemistry 22 (2012)4611–4614.

18] S. Shi, L. Liu, C. Ouyang, D.-s. Wang, Z. Wang, L. Chen, X. Huang, Enhancement ofelectronic conductivity of LiFePO4 by Cr doping and its identification by first-principles calculations, Physical Review B 68 (2003) 195108.

19] T. Adschiri, Y.W. Lee, M. Goto, S. Takami, Green materials synthesis with super-critical water, Green Chemistry 13 (2011) 1380–1390.

20] T. Adschiri, Y. Hakuta, K. Sue, K. Arai, Hydrothermal synthesis of metal oxidenanoparticles at supercritical conditions, J. Nanoparticle Research 3 (2001)227–235.

21] T. Adschiri, K. Kanazawa, K. Arai, Rapid and continuous hydrothermal synthesisof boehmite particles in subcritical and supercritical water, J. The AmericanCeramic Society 75 (1992) 2615–2618.

22] K. Sue, M. Suzuki, K. Arai, T. Ohashi, H. Ura, K. Matsui, Y. Hakuta, H. Hayashi, M.Watanabed, T. Hiakia, Size-controlled synthesis of metal oxide nanoparticleswith a flow-through supercritical water method, Green Chemistry 8 (2006)634–638.

23] T. Adschiri, Y. Hakuta, K. Arai, Hydrothermal synthesis of metal oxide fine par-ticles at supercritical conditions, Industrial & Engineering Chemistry Research39 (2000) 4901–4907.

24] K. Kanamura, A. Goto, R.Y. Ho, T. Umegaki, K. Toyoshima, K.-i. Okada, Y. Hakuta,T. Adschiri, K. Arai, Preparation and electrochemical characterization of LiCoO2

particles prepared by supercritical water synthesis, Electrochemical and SolidState Letters 3 (2000) 256–258.

25] Y.H. Shin, S.-M. Koo, D.S. Kim, Y.-H. Lee, B. Veriansyah, J. Kim, Y.-W. Lee, Continu-ous hydrothermal synthesis of HT-LiCoO2 in supercritical water, J. SupercriticalFluids 50 (2009) 250–256.

26] K. Kanamura, K. Dokko, T. Kaizawa, Synthesis of spinel LiMn2O4 by ahydrothermal process in supercritical water with heat-treatment, J. The Elec-trochemical Society 152 (2005) A391–A395.

27] J.-W. Lee, J.-H. Lee, T.T. Viet, J.-Y. Lee, J.-S. Kim, C.-H. Lee, Synthesis ofLiNi1/3Co1/3Mn1/3O2 cathode materials by using a supercritical water methodin a batch reactor, Electrochimica Acta 55 (2010) 3015–3021.

28] J. Lee, A.S. Teja, Characteristics of lithium iron phosphate (LiFePO4) particlessynthesized in subcritical and supercritical water, J. Supercritical Fluids 35(2005) 83–90.

29] C. Xu, J. Lee, A.S. Teja, Continuous hydrothermal synthesis of lithium iron phos-phate particles in subcritical and supercritical water, J. Supercritical Fluids 44(2008) 92–97.

30] J. Lee, A.S. Teja, Synthesis of LiFePO4 micro and nanoparticles in supercriticalwater, Materials Letters 60 (2006) 2105–2109.

[

[

cal Fluids 73 (2013) 70– 79 79

31] A. Aimable, D. Aymes, F. Bernard, F.L. Cras, Characteristics of LiFePO4 obtainedthrough a one step continuous hydrothermal synthesis process working insupercritical water, Solid State Ionics 180 (2009) 861–866.

32] S.A. Hong, S.J. Kim, J. Kim, K.Y. Chung, B.W. Cho, J.W. Kang, Small capacity decayof lithium iron phosphate (LiFePO(4)) synthesized continuously in supercriticalwater: comparison with solid-state method, J. Supercritical Fluids 55 (2011)1027–1037.

33] S.-A. Hong, S.J. Kim, J. Kim, B.G. Lee, K.Y. Chung, Y.-W. Lee, Carbon coatingon lithium iron phosphate (LiFePO4): comparison between continuous super-critical hydrothermal method and solid-state method, Chemical Engineering J.198–199 (2012) 318–326.

34] A. Nugroho, S.J. Kim, K.Y. Chung, J. Kim, Synthesis of Li4Ti5O12 in supercriti-cal water for Li-ion batteries: reaction mechanism and high-rate performance,Electrochimica Acta 78 (2012) 623–632.

35] A. Nugroho, S.J. Kim, K.Y. Chung, B.-W. Cho, Y.-W. Lee, J. Kim, Facile synthesis ofnanosized Li4Ti5O12 in supercritical water, Electrochemistry Communications13 (2011) 650–653.

36] A. Laumann, M. Bremholm, P. Hald, M. Holzapfel, K.T. Fehr, B.B. Iversen, Rapidgreen continuous flow supercritical synthesis of high performance Li4Ti5O12

nanocrystals for Li ion battery applications, J. The Electrochemical Society 159(2012) A166–A171.

37] K. Sue, T. Sato, S. Kawasaki, Y. Takebayashi, S. Yoda, T. Furuya, T. Hiaki,Continuous hydrothermal synthesis of Fe2O3 nanoparticles using a cen-tral collision-type micromixer for rapid and homogeneous nucleation at673 K and 30 MPa, Industrial & Engineering Chemistry Research 49 (2010)8841–8846.

38] K. Sue, S. Kawasaki, M. Suzuki, Y. Hakuta, H. Hayashi, K. Arai, Y. Takebayashi,S. Yoda, T. Furuya, Continuous hydrothermal synthesis of Fe2O3, NiO, and CuOnanoparticles by superrapid heating using a T-type micro mixer at 673 K and30 MPa, Chemical Engineering J. 166 (2011) 947–953.

39] A. Aimable, H. Muhr, C. Gentric, F. Bernard, F.L. Cras, D. Aymes, Continu-ous hydrothermal synthesis of inorganic nanopowders in supercritical water:towards a better control of the process, Powder Technology 190 (2009)99–106.

40] Y. Wakashima, A. Suzuki, S.-i. Kawasaki, K. Matsui, y. Hakuta, Development of anew swirling micro mixer for continuous hydrothermal synthesis of nano-sizeparticles, J. Chemical Engineering of Japan 40 (2007) 622–629.

41] S.I. Kawasaki, K. Sue, R. Ookawara, Y. Wakashima, A. Suzuki, Y. Hakuta, K. Arai,Engineering study of continuous supercritical hydrothermal method using aT-shaped mixer: experimental synthesis of NiO nanoparticles and CFD simula-tion, J. Supercritical Fluids 54 (2010) 96–102.

42] E. Lester, P. Blood, J. Denyer, D. Giddings, B. Azzopardi, M. Poliakoff, Reactionengineering: the supercritical water hydrothermal synthesis of nano-particles,J. Supercritical Fluids 37 (2006) 209–214.

43] J. Kim, Y.S. Park, B. Veriansyah, J.D. Kim, Y.W. Lee, Continuous syn-thesis of surface-modified metal oxide nanoparticles using supercriticalmethanol for highly stabilized nanofluids, Chemistry of Materials 20 (2008)6301–6303.

44] H. Choi, B. Veriansyah, J. Kim, J.D. Kim, J.W. Kang, Continuous synthesis ofmetal nanoparticles in supercritical methanol, J. Supercritical Fluids 52 (2010)285–291.

45] B. Veriansyah, J.D. Kim, B.K. Min, Y.H. Shin, Y.W. Lee, J. Kim, Continuous syn-thesis of surface-modified zinc oxide nanoparticles in supercritical methanol,J. Supercritical Fluids 52 (2010) 76–83.

46] B. Veriansyah, J.D. Kim, B.K. Min, J. Kim, Continuous synthesis of mag-netite nanoparticles in supercritical methanol, Materials Letters 64 (2010)2197–2200.

47] K. Sue, M. Aoki, T. Sato, D. Nishio-Hamane, S.-i. Kawasaki, Y. Hakuta, Y. Take-bayashi, S. Yoda, T. Furuy, T. Sato, T. Hiaki, Continuous hydrothermal synthesisof nickel ferrite nanoparticles using a central collision-type micriomixer: effectof temperature, residence time, metal salt molarity, and NaOH addition on con-version, particle size, and crystal phase, Industrial & Engineering ChemistryResearch 50 (2011) 9625–9631.

48] L.J. Cote, A.S. Teja, A.P. Wilkinson, Z.J. Zhang, Continuous hydrothermal synthe-sis of CoFe2O4 nanoparticles, Fluid Phase Equilibria 210 (2003) 307–317.

49] Y. Hakuta, H. Ura, H. Hayashi, K. Arai, Effects of hydrothermal synthetic condi-tions on the particle size of �-AlO(OH) in sub and supercritical water using aflow reaction system, Materials Chemistry and Physics 93 (2005) 466–472.

50] K. Sue, K. Murata, K. Kimura, K. Arai, Continuous synthesis of zinc oxidenanoparticles in supercritical water, Green Chemistry 5 (2003) 659–662.

51] B. Ellis, W.H. Kan, W.R.M. Makahnouk, L.F. Nazar, Synthesis of nanocrystals andmorphology control of hydrothermally prepared LiFePO4, J. Materials Chem-istry 17 (2007) 3248–3254.

52] J. Barker, M.Y. Saidi, J.L. Swoyer, Lithium iron (II) phospho-olivines preparedby a novel carbothermal reduction method, Electrochemical and Solid State

Letters 6 (2003) A53–A55.

53] B.V. L’vov, Mechanism of carbothermal reduction of iron, cobalt, nickel andcopper oxides, Thermochimica Acta 360 (2000) 109–120.

54] H.-S. Kim, B.-W. Cho, W.-I. Cho, Cycling performance of LiFePO4 cathode mate-rial for lithium secondary batteries, J. Power Sources 132 (2004) 235–239.