Self-assembly of LiFePO 4 nanodendrites in a novel system of ethylene glycol–water Fei Teng a,c , Sunand Santhanagopalan a , Anjana Asthana a , Xiaobao Geng a , Sun-il Mho b , Reza Shahbazian-Yassar a , Dennis Desheng Meng a,n a Department of Mechanical Engineering-Engineering Mechanics, Michigan Technological University, Houghton, MI 49931, USA b Division of Energy System research, Ajou University, Suwon 443-749, Korea c School of Environmental Science and Engineering, Nanjing University of Information Science and Technology Nanjing 210044, P.R. China article info Article history: Received 24 March 2010 Received in revised form 9 July 2010 Accepted 2 September 2010 Communicated by M. Schieber Keywords: A1. Crystal morphology A1. Crystal structure A1. Nanostructures A2. Self-assembly B1. LiFePO 4 abstract In this work, a novel system of ethylene glycol/water (EG/W) was employed to synthesize LiFePO 4 , in which dodecyl benzene sulphonic acid sodium (SDBS) was used as soft template to control particle morphology. The samples were characterized by X-ray diffractometer (XRD), field emission scanning electron microscopy (FE-SEM), high resolution transmission electron microscopy (HRTEM), selected area electron diffraction (SAED) and energy dispersive X-ray spectroscopy (EDX). The LiFePO 4 sample obtained by the reported method displays interesting hierarchical nanostructure (i.e. nanodendrites), which was constructed by nanorods of 3–5 mm in length and 50 nm in diameter. The EG/W ratio, amount of SDBS added, hydrothermal temperature and duration played important roles in the assembly of the hierarchical nanostructures. A formation mechanism was proposed and experimentally verified. It is concluded that the nanodendrites were formed due to the end-to-end self-assembly of nanorods. Compared to previously reported methods, the reported approach shows obvious advantages of one- step synthesis, environmental friendliness and low cost, to name a few. The nanodendrites as a cathode material have a higher capacity, compared with the other samples. & 2010 Published by Elsevier B.V. 1. Introduction Since the pioneering work of Padhi et al. [1], LiFePO 4 has been intensively investigated as one of the most promising cathode materials in rechargeable lithium batteries [2,3]. The main advan- tages of LiFePO 4 include flat voltage plateau (3.4 V versus Li + /Li), moderate theoretical specific capacity (170 mAh g 1 ), high thermal stability, long cycling life, environmental friendliness and abundance of iron (Fe) resources in nature. Moreover, LiFePO 4 is considered to be much safer than other cathode materials (e.g., LiCoO 2 and LiMn 2 O 4 ) due to its outstanding stability upon overcharging and overdischarging [1–3]. For example, metastable LiCoO 2 can easily lose oxygen while overcharging, which increases the probability of overheating and electrolyte decomposition. In contrast, due to the strong P–O covalent bonds in PO 34 polyanion, phospho-olivine LiFePO 4 has a very stable three-dimensional framework, which greatly reduces the risk of oxygen liberation. Hence, LiFePO 4 is expected to be the most promising cathode material for large-size lithium-ion batteries in electrical vehicles (EVs) and hybrid electric vehicles (HEVs) that demand both fast charging and strict safety regulation [1,4]. However, bulk LiFePO 4 electrodes are limited on its rate capability, due to its intrinsic low electrical conductivity and limited Li ion diffusion rate [2,3,5–7]. Its electrical conductivity (10 9 –10 10 S cm 1 ) is by several orders of magnitude lower than that (e.g., ca. 10 3 S cm 1 ) of LiCoO 2 cathode materials, which prevents achievement of full theoretical capacity (170 mAh g 1 ) at very high rates [3,7]. To date, most attempts have been carried out to overcome the electronic and lithium-diffusion limitations, such as coating of carbon [3,7,8], cationic supervalent substitution [3,7,9,10] and addition of conductive materials (Cu, Ag, carbon black) [11–13]. During the last decade, it has also been demonstrated that size reduction of LiFePO 4 micro- or nanoparticles can effectively enhance Li diffusion rate and significantly improve performances [3,14]. Another significant challenge is the large-scale synthesis of high- purity LiFePO 4 nanoparticles with uniform morphology, which is essential for their commercial applications. To date, various methods have been proposed to synthesize LiFePO 4 , such as solid-state reaction [15], sol–gel [16], hydrothermal [17,18], co-precipitation [7], vapor deposition [8], microwave [19], spraying technology [20], to name a few. Of these, hydrothermal synthesis is an effective method to obtain well-crystallized materials with well-defined morphologies, where no additional high-temperature annealing is needed [17,21]. To date, synthesis of di-element materials with various morphologies by a hydrothermal route has been a fairly common practice. Nevertheless, morphosynthesis of multi-element LiFePO 4 represents significant challenges and thus has been reported Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/jcrysgro Journal of Crystal Growth 0022-0248/$ - see front matter & 2010 Published by Elsevier B.V. doi:10.1016/j.jcrysgro.2010.09.005 n Corresponding author. Tel./fax: + 1 906 4873551. E-mail address: [email protected] (D.D. Meng). Please cite this article as: F. Teng, et al., J. Crystal Growth (2010), doi:10.1016/j.jcrysgro.2010.09.005 Journal of Crystal Growth ] (]]]]) ]]]–]]]
Self-assembly of LiFePO4 nanodendrites in a novel systemof ethylene glycol–water
Fei Teng a,c, Sunand Santhanagopalan a, Anjana Asthana a, Xiaobao Geng a, Sun-il Mho b,Reza Shahbazian-Yassar a, Dennis Desheng Meng a,n
a Department of Mechanical Engineering-Engineering Mechanics, Michigan Technological University, Houghton, MI 49931, USAb Division of Energy System research, Ajou University, Suwon 443-749, Koreac School of Environmental Science and Engineering, Nanjing University of Information Science and Technology Nanjing 210044, P.R. China
a r t i c l e i n f o
Article history:
Received 24 March 2010
Received in revised form
9 July 2010
Accepted 2 September 2010
Communicated by M. Schieberarea electron diffraction (SAED) and energy dispersive X-ray spectroscopy (EDX). The LiFePO4 sample
In this work, a novel system of ethylene glycol/water (EG/W) was employed to synthesize LiFePO4, in
which dodecyl benzene sulphonic acid sodium (SDBS) was used as soft template to control particle
morphology. The samples were characterized by X-ray diffractometer (XRD), field emission scanning
electron microscopy (FE-SEM), high resolution transmission electron microscopy (HRTEM), selected
obtained by the reported method displays interesting hierarchical nanostructure (i.e. nanodendrites),
which was constructed by nanorods of 3–5 mm in length and �50 nm in diameter. The EG/W ratio,
amount of SDBS added, hydrothermal temperature and duration played important roles in the assembly
of the hierarchical nanostructures. A formation mechanism was proposed and experimentally verified.
It is concluded that the nanodendrites were formed due to the end-to-end self-assembly of nanorods.
Compared to previously reported methods, the reported approach shows obvious advantages of one-
step synthesis, environmental friendliness and low cost, to name a few. The nanodendrites as a cathode
material have a higher capacity, compared with the other samples.
& 2010 Published by Elsevier B.V.
1. Introduction
Since the pioneering work of Padhi et al. [1], LiFePO4 has beenintensively investigated as one of the most promising cathodematerials in rechargeable lithium batteries [2,3]. The main advan-tages of LiFePO4 include flat voltage plateau (3.4 V versus Li+/Li),moderate theoretical specific capacity (170 mAh g�1), high thermalstability, long cycling life, environmental friendliness and abundanceof iron (Fe) resources in nature. Moreover, LiFePO4 is considered tobe much safer than other cathode materials (e.g., LiCoO2 andLiMn2O4) due to its outstanding stability upon overcharging andoverdischarging [1–3]. For example, metastable LiCoO2 can easilylose oxygen while overcharging, which increases the probability ofoverheating and electrolyte decomposition. In contrast, due to thestrong P–O covalent bonds in PO3�
4 polyanion, phospho-olivineLiFePO4 has a very stable three-dimensional framework, whichgreatly reduces the risk of oxygen liberation. Hence, LiFePO4 isexpected to be the most promising cathode material for large-sizelithium-ion batteries in electrical vehicles (EVs) and hybrid electricvehicles (HEVs) that demand both fast charging and strict safetyregulation [1,4]. However, bulk LiFePO4 electrodes are limited on its
Elsevier B.V.
J. Crystal Growth (2010), do
rate capability, due to its intrinsic low electrical conductivity andlimited Li ion diffusion rate [2,3,5–7]. Its electrical conductivity(10�9–10�10 S cm�1) is by several orders of magnitude lower thanthat (e.g., ca. 10�3 S cm�1) of LiCoO2 cathode materials, whichprevents achievement of full theoretical capacity (170 mAh g�1) atvery high rates [3,7]. To date, most attempts have been carried out toovercome the electronic and lithium-diffusion limitations, such ascoating of carbon [3,7,8], cationic supervalent substitution [3,7,9,10]and addition of conductive materials (Cu, Ag, carbon black) [11–13].During the last decade, it has also been demonstrated that sizereduction of LiFePO4 micro- or nanoparticles can effectively enhanceLi diffusion rate and significantly improve performances [3,14].Another significant challenge is the large-scale synthesis of high-purity LiFePO4 nanoparticles with uniform morphology, which isessential for their commercial applications. To date, various methodshave been proposed to synthesize LiFePO4, such as solid-statereaction [15], sol–gel [16], hydrothermal [17,18], co-precipitation[7], vapor deposition [8], microwave [19], spraying technology [20],to name a few. Of these, hydrothermal synthesis is an effectivemethod to obtain well-crystallized materials with well-definedmorphologies, where no additional high-temperature annealing isneeded [17,21]. To date, synthesis of di-element materials withvarious morphologies by a hydrothermal route has been a fairlycommon practice. Nevertheless, morphosynthesis of multi-elementLiFePO4 represents significant challenges and thus has been reported
Fig. 1. XRD patterns of the as-prepared samples at different hydrothermal
temperatures corresponding to those in Table 1. (A) scanning in the range of
15–701; (B) slow scanning in the range of 26–321: (a) S1-1; (b) S1-2; (c) S1-3;
(d) S1-4.
F. Teng et al. / Journal of Crystal Growth ] (]]]]) ]]]–]]]2
only scarcely. This situation has been attributed to the fact that thedifferent precursor chemicals generally have different reactivities,which is not favorable for crystal growth or formation of pureproducts. In the past years, Nazar et al. investigated the influence ofhydrothermal conditions and additives on the morphology ofLiFePO4 crystals [18]. Chen et al. [22] prepared the large diamond-shaped LiFePO4 sample by the hydrothermal method. Sides et al.[23] synthesized LiFePO4 nanofibers with polycarbonate membranestemplate. They reported that the nanofibers showed an excellentrate capacity because the nanofiber morphology mitigates theproblem of slow Li ion-transport in solid state. Vittal et al. alsosynthesized LiFePO4 nanoplates with high rate capability [24]. Theirresults showed that particle morphology or architecture has a greatinfluence on the performances of materials [25]. Although signifi-cant efforts have to be made to synthesize LiFePO4 nanostructures,designing or controling new and advanced nanostructures withenhanced performances is always of most interest to materialists.Recently, Rangappa et al. [26] synthesized LiFePO4 flowers using thesolvothermal method. However, the preparation was performed at avery high temperature (400 1C) and pressure (40 MPa), which limitsits practical production. Yang et al. [27] synthesized LiFePO4
dumbbell-like microstructures composed of nanoplates via asolvothermal route, which showed an excellent cycling stability. Intheir work, nevertheless, expensive benzyl alcohol and LiI were used,and a long reaction time (48 h) was required. Hence, the preparationis costly and time-consuming. Much effort is still needed to developa more economically efficient route to synthesize LiFePO4 with well-defined morphology. Although the hydrothermal method has shownits potential to provide such a route, it has not been reported as afeasible approach to synthesize LiFePO4 nanodendrites.
In this work, a novel system of ethylene glycol/water (EG/W)was developed to synthesize LiFePO4 nanodendrites under hydro-thermal conditions. Lithium hydroxide was used as a precursor anddodecyl benzene sulphonic acid sodium (SDBS) was used as a softtemplate to control crystal growth. LiFePO4 samples have beenprepared under various synthesis conditions to reveal the influ-ences of EG/W volumetric ratio, amount of SDBS added, hydro-thermal temperature and time. The samples were characterized byXRD, FE-SEM, HRTEM, ED and EDX. A formation mechanism ofLiFePO4 nanodendrites was proposed based on the time-dependentresults. The reported approach shows significant advantages overthe state of the art, such as one-pot synthesis, environmentalfriendliness and low cost, which are mainly attributed to theemployed inexpensive precursors and solvent, as well as the short
Table 1Effects of hydrothermal temperature and duration on particle shape, crystal structure and sizea of the sample.
a Crystal size calculated by the Sherrer equation basing on [0 2 0] plane.b Ethylene glycol/deionized water volumetric ratio.c Molar ratio; SDBS stands for dodecyl benzene sulphonic acid sodium.d Observed by FE-SEM or HRTEM.
Please cite this article as: F. Teng, et al., J. Crystal Growth (2010), doi:10.1016/j.jcrysgro.2010.09.005
Fig. 2. FE-SEM micrographs of the as-prepared samples corresponding to those in Table 1: (a) and (b) S1-2 at different magnifications, (c) S1-3 and (d) S1-4.
20
(022
)
(020
)
(201
)
(430
)(3
31)
(412
)
(113
)
(222
)(4
21)
(112
)
(221
)(4
10)
(121
)(3
11)
(301
)(2
11)
(111
)(0
11)
(210
)(101
)
(200
)
* * # *##
Inte
nsity
/a.u
.
2 Theta/degree
(a)
(b)
(c)
(d)
(e)
*: Unknown#: Fe4(PO4)3(OH)3
30 40 50 60 70
Fig. 3. XRD patterns of the as-prepared samples at different hydrothermal
durations correspond to those in Table 1: (a) S1-5, (b) S1-6, (c) S1-2, (d) S1-7
and (e) S-8.
F. Teng et al. / Journal of Crystal Growth ] (]]]]) ]]]–]]] 3
process time (�6 h). The novel nanodendrites are expected topresent excellent electrochemical properties compared to othernanostructures, e.g., near-spherical nanoparticles.
2. Experimental section
2.1. Chemicals
In this work, all chemicals were used as purchased withoutfurther purification. Deionized (DI) water treated in-house by anion-exchange system was used in the experiment. FeCl2 �7H2O,LiOH, H3PO4 (85 wt% aqueous solution), L(+)-ascorbic acid,ethylene glycol (EG) and dodecyl benzene sulphonic acid sodium(SDBS, C18H29SO3Na) were purchased from Sigma Aldrich.
Please cite this article as: F. Teng, et al., J. Crystal Growth (2010), do
2.2. Preparation of samples
The LiFePO4 nanodendrites were prepared by a simple hydro-thermal process, during which L-ascorbic acid was added as a mildreducing agent to prevent the oxidation of Fe(II). Typically, theappropriate quantities of LiOH, FeCl2 �7H2O, H3PO4, L-ascorbic acidand SDBS with the molar ratios of 3:1:1:3:1 were added into 35 mLof EG/W mixture (1/1 volumetric ratio). After intensive magneticstirring for 1 h at room temperature, a homogeneous solution wasformed. The solution was then transferred into a 50 mL Teflons
-lined stainless steel autoclave, which was heated to 160 1C andkept at this temperature for 6 h. After being cooled naturally toroom temperature, the product was centrifuged, washed at leastsix times with DI water and ethanol, and finally dried at 80 1C in avacuum for 24 h. In order to investigate the effects of preparationconditions on the samples, several experimental parameters, suchas the volumetric ratio of EG to W, the amount of SDBS added,hydrothermal time and temperature, were varied.
2.3. Characterization
Scanning electron microscopy (SEM) images were taken with aHitachi S-4700 field emission scanning electron microscope(FE-SEM). Before FE-SEM inspection, the samples were coatedwith 5-nm-thick platinum/palladium layer by direct current(DC) sputtering. The acceleration voltage was 15 keV, and theacceleration current was 1.2 nA. The morphology, crystallineproperties, surface structure and element composition of thesamples were determined by using high-resolution transmissionelectron microscopy (HRTEM). We employed a JEOL JEM-4000FXsystem equipped with electron diffraction and energy dispersiveX-ray spectroscopy (EDX) attachments with an accelerationvoltage of 200 kV. The powders were first ultrasonically dispersedin ethanol, and then deposited on a thin amorphous carbon filmsupported by a copper grid. The crystal structures of the sampleswere characterized by an X-ray powder diffractometer (XRD,Rigaku D/MAX-RB), using graphite monochromatized Cu Ka
Fig. 4. HRTEM micrographs of the as-prepared samples at different reaction times
corresponding to those in Table 1: (a) t¼2 h (S1-5), (b) t¼4 h (S1-6), (c) t¼6 h
(S1-2), (d) t¼12 h (S1-7), (e) the tip of an individual particle for sample S1-2,
(f) the inset of SAED image, (g) lattice fringe for sample S1-2 and (h) energy
dispersive X-ray spectra (EDX) for sample S1-2.
F. Teng et al. / Journal of Crystal Growth ] (]]]]) ]]]–]]]4
radiation (l¼0.154 nm), operating at 40 kV and 50 mA. The XRDpatterns were obtained in the range of 15–701 (2y) at a scanningrate of 51 min�1. A nitrogen adsorption isotherm was performedat 77 K and o10�4 bar on a Micromeritics ASAP2010 gasadsorption analyzer. Each sample was degassed at 200 1C for 5 hbefore the measurement. Surface area and pore size distributionwere calculated by the BET (Brunauer–Emmett–Teller) method.
2.4. Electrochemical measurements
Electrochemical measurements were performed using a CR2032coin cell assembled in an argon-filled glove box. The working
Please cite this article as: F. Teng, et al., J. Crystal Growth (2010), do
electrode slurry was composed of 75 wt% LiFePO4 powder, 15 wt%acetylene black, 10 wt% polyvinylidene fluoride (PVDF) binder.N-methyl-2-pyrrolidinon (NMP) was used as solvent. The slurrywas cast onto the aluminum current collector and achieved aloading of �1 mg cm�2. The electrode was then dried overnightunder vacuum at 120 1C. A Teflon Celgard separator (#2400) wasused to separate the working electrode and a lithium foil counter-electrode. The electrolyte consisted of a solution of 1 M LiPF6 inethylene carbonate (EC)/dimethyl carbonate (DMC)/diethyl carbo-nate (DEC) (1:1:1, in wt%) obtained from Ferro Corporation.Charge/discharge test of the assembled coin cell was performedon an Arbin BT2000 system in the voltage range of 2.0–4.2 V atdifferent C rates of 0.1C–5C (C¼170 mA g�1) at 25 1C.
3. Results and discuss
3.1. Influences of hydrothermal temperature and duration
First, we investigated the influence of hydrothermal tempera-ture on the sample (Table 1 S1-1 to S1-4). The hydrothermaltemperature was varied from 140 to 160, 180 and 200 1C, whilethe other processing parameters were maintained the same. Fig. 1gives the typical XRD patterns of the as-synthesized samples.After being processed at 140 1C for 6 h, no single-phase LiFePO4
sample was obtained (S1-1). Instead, the sample containedsignificant amounts of impurity crystals (e.g., Li3PO4 andFe4(PO4)3(OH)3). It could be assumed that the higher energy isnecessary for the formation of LiFePO4 crystals. Above 160 1C, thediffraction peaks of all the samples (S1-2, S1-3 and S1-4 inFig. 1(A)) can be well indexed to single-phase LiFePO4 with anorthorhombic olivine structure (JCPDS card no. 81-1173). More-over, the diffraction peaks become stronger and narrower withthe increase of temperature. The mean crystallite size (D020) wasfurther calculated using Scherrer’s equation from the [0 2 0]diffraction peak (Fig. 1(B)). The mean D020 values for S1-2, S1-3and S1-4 are 25.5, 37, and 42 nm, respectively. It is obvious thatthe crystals have grown larger with the increase in hydrothermaltemperature. These results show that the reaction temperaturehas a significant influence on crystal growth. Further, themorphologies of the as-prepared samples (S1-2, S1-3, and S1-4)were characterized by FE-SEM. Interestingly, the as-preparedsample (S1-2) at 160 1C displays a hierarchical structure(i.e. nanodendrites), as illustrated in Fig. 2(a) and (b). Whenobserved at a low magnification (Fig. 2(a)), the particles displayedfairly uniform morphology, and each particle features a nanoden-drite of 4–5 mm in length. As revealed in Fig. 2(b), the individualnanodendrite consists of nanorods of 100 nm in diameter and2–5 mm in length. These nanorods seem to be attached end-to-endto form an ordered architecture. This novel hierarchical architec-ture of LiFePO4 nanorods has not been reported previously. Athigher temperatures (180 and 200 1C), however, the as-preparedsamples display irregular morphologies (Fig. 2(c) and (d)). It isobvious that the hydrothermal temperature has a significantinfluence on the morphology and structure of the products. Itcould be concluded that there exists a critical temperature at whichthe nucleation and growth of crystals are too fast to form crystalswith well-defined morphology. It is clear that the hydrothermaltemperature played a key role in the formation of high crystallinityLiFePO4 crystals. Our results have shown that an appropriatehydrothermal temperature (160 1C) can be chosen to synthesizeLiFePO4 nanodendrites. This mild temperature of the reportedprocess is considered to be a major advantage over existingmethods employing higher processing temperatures.
Hydrothermal time was then varied from 2, 6, 12 and 24 h toinvestigate the effect of reaction time while the hydrothermal
a Volumetric ratio.b Molar ratio.c Observed by FE-SEM.
20
#
# #
##
#
#**
Inte
nsity
/a.u
.
2 Theta/degree
(a)
(b)
(c)
(d)
*: Li3PO4
#: Fe4(PO4)3(OH)3
#*
30 40 50 60 70
Fig. 5. XRD patterns of the as-prepared samples at different volumetric ratios of
EG/W corresponding to those in Table 2: (a) S2-1, (b) S2-3, (c) S2-4 and (d) S2-5.
F. Teng et al. / Journal of Crystal Growth ] (]]]]) ]]]–]]] 5
temperature was kept at 160 1C (Table 1, S1-2, and S1-5 to S1-8).Fig. 3 shows XRD patterns of the as-prepared samples. After beingprocessed for 2 h, no LiFePO4 was evidenced, but the weak peaks ofFe4(PO4)3(OH)3 and unknown phases can be observed (Fig. 3(a)).When the reaction time was extended to 4 h, the peaks of bothLi3Fe2(PO4)3 and Li3PO4 disappeared (Fig. 3(b)). At the same time,all the diffraction peaks, although relatively weak, can be wellindexed to single-phase LiFePO4 with an orthorhombic olivinestructure. When the reaction time was extended to 6 h, all thediffraction peaks of the sample turned out to be strong, and can bewell indexed to high-purity LiFePO4 structure (Sample S1-2).Further increasing reaction time to 12 and 24 h results in strongerall diffraction peaks for samples S1-7 and S1-8. The crystallite size(D020) was calculated using Scherrer’s equation (D¼0.9l/b cos y)from the full-width-at-half-maximum (b) of the strong and well-resolved diffraction peak [0 2 0]. The calculated mean D020 valuesfor S1-6, S1-2, S1-7 and S1-8 are 18.8, 25.5, 27.1 and 27.5 nm,respectively. The results showed that after the processing time waslonger than 6 h, the crystals grew at a fairly smaller extent. It istherefore concluded that the crystal growth slowed down after 6 h.An even longer processing time will have minor influence on thecrystal size and morphology of the reported LiFePO4 nanodendrites.Further, HRTEM was performed to investigate the variation ofLiFePO4 particle morphology with processing time, as shown inFig. 4. It is revealed that after being processed for 2 h at 160 1C, thesample (S1-5) was composed of 10-nm-large nanoparticles(Fig. 4(a)). On the other hand, when the processing time isextended to 4 h, the sample obtained (S1-6) turned out to beLiFePO4 nanorods that are 100 nm in length and 10–20 nm indiameter (Fig. 4(b)). After being processed for 6 h, the HRTEM
Please cite this article as: F. Teng, et al., J. Crystal Growth (2010), do
images further confirm that the as-obtained sample consists ofmany well-dispersed nanodendrites with uniform morphology.The length of those structures is again confirmed as of 4–5 mm(Fig. 4(c)). The results consistently agree with the FE-SEM imagespresented in Fig. 2(a) and (b). It is worth mentioning that thesenanodendrites are sufficiently stable to withstand ultrasonictreatment for 30 min (or even longer) without being broken intofragments. We also observed that after being processed for 12 or24 h, the nanodendrite architectures maintained their morpholo-gies (Fig. 4(d)). The TEM image for sample S1-8 (24 h) is not shownherein. To exemplify the morphology of the sample obtained, thehigh-magnification image of an individual particle is shown inFig. 4(d). The spiky structure on the edge of the particle indicatesthat the particles are composed of nanorods with a diameter ofabout 100 nm (Fig. 4(e)). The inset of ED patterns (Fig. 4(f))indicates the single-crystalline nature of the nanorods. A HRTEMimage (Fig. 4(g)) taken on the tip of an individual nanorod displaysclear crystal lattices with d-spacing of 0.29 nm, corresponding tothe (0 2 0) plane of orthorhombic LiFePO4 crystals. The resultfurther confirmed the single-crystalline nature of the nanorods,and suggested that the crystal preferentially grow along the {0 2 0}direction. Fig. 4(h) provides the EDX spectra of the sample (S1-2).The peaks of copper are obviously caused by the copper grid, whichis the carrier of the HRTEM samples. The peaks of Fe and P can beobserved clearly, but the peaks of lithium did not appear in EDXspectra due to the light mass of lithium. The above time-dependentexperiments suggest that at the beginning stage, the nucleationprocess takes place under hydrothermal conditions, leading to theformation of nanoparticles; then the nanoparticles grow intonanorods through a crystallization–dissolvation––recrystallizationprocess. Eventually, the nanorods organize into the hierarchicalstructures through an end-to-end self-assembly process. Extensivefuture research is still needed to clarify the details of the self-assembly process.
3.2. Influence of reaction medium composition
We have also investigated the effect of reaction medium on thesamples. EG/W volumetric ratio was varied at 0/1, 1/1, 2/1, 4/1 and1/0 while other preparation parameters were maintained the same(Table 2). The typical XRD patterns and FE-SEM images of the as-prepared samples are shown in Figs. 5 and 6, respectively. At theEG/W ratios of 0/1, the sample obtained is mainly composed ofLiFePO4 crystals accompanied by a very small amount of Fe4(PO4)
3(OH)3 impurity crystals (Fig. 5(a)). At the EG/W ratios of 1/1 and2/1, single-phase LiFePO4 crystals were obtained, while no otherimpurity crystals were observed (Fig. 5(b), S1-2 in Fig. 2). At theEG/W volumetric ratio of 0/1, the obtained sample is composed ofplates with a thickness of 100–150 nm and a width of 1 mm (S2-1in Fig. 6). The nanodendrite structure was obtained at an EG/Wvolumetric ratio of 1/1, as shown in Fig. 2. When the volumetricratio of EG/W was changed to 2/1, the sample obtained consists of
Fig. 6. FE-SEM images of the as-prepared samples at different volumetric ratios of EG/W corresponding to those in Table 2: (a) S2-1, (b) S2-3, (c) S2-4 and (d) S2-5.
Table 3Effect of the SDBS addition amount on particle shape and crystal structure of the sample.
S3-1 1/1 0/1 160/6 LiFePO4 Large spheres+irregular particles
S1-2 1/1 1/1 160/6 LiFePO4 Nanodendrites
S3-3 1/1 2/1 160/6 LiFePO4 Rods+dumbbells
S3-4 1/1 4/1 160/6 LiFePO4 Large bones
a Volumetric ratio.b Molar ratio.c Observed by FE-SEM.
F. Teng et al. / Journal of Crystal Growth ] (]]]]) ]]]–]]]6
large spindle-shaped particles with a length of 2 mm and a centerdiameter of 1 mm (S2-3 in Fig. 6). When the volumetric ratio of EGto W was further increased to 4/1 and 1/0, high-purity LiFePO4
crystals were not obtained. Instead, the impurity crystals (Li3PO4,FeFe3(PO4)3(OH)3 and unknown phases) started to appear in thesesamples. FE-SEM images show that both samples consist of largerods and irregular particles (S2-4 and S2-5 in Fig. 6). It is obviousthat reacting medium composition has a significant influence onphase composition and morphology of the samples. Two aspects ofthe new mixture medium are assumed to be very important in thisprocess. First of all, a high enough solubility of the precursorchemicals in the EG/W medium needs to be guaranteed to ensurethe formation of proper LiFePO4 crystal structure. As the EG contentin the mixture increases, the solubility of the precursor chemicalsdecreases. As a result, the precursor chemicals cannot be mixedhomogeneously at higher EG content, which does not favor theformation of pure LiFePO4 crystals. On the other hand, it is wellknown that the viscosity of EG (Z¼21 mPa s, 20 1C) is much higherthan that of water (Z¼1.0087�10�3 mPa s, 20 1C). The mobility or
Please cite this article as: F. Teng, et al., J. Crystal Growth (2010), do
reactivity of ions in solvent has an important influence on thecrystal growth. Too higher EG content may lead to very slowmobility or reactivity of ions in the medium, which does not favorto form the final product. At an appropriate content of EG (i.e. 1/1volumetric ratio of EG to W), the reactivity of the precursors maymatch one another, which does favor to form the final product.Although those two factors can explain our experimental resultswell, further investigations are still needed to confirm ourhypothesis and reveal the fundamental physical and chemicalinteractions in our process.
3.3. Influence of the amount of surfactant added
To investigate the influence of the added amount of SDBS onthe self-assembly behavior of LiFePO4 nanostructures, thisparticular processing parameter (SDBS amount) was changedwhile others were maintained the same, as shown in Table 3.Fig. 7 indicates high-purity LiFePO4 structures of the as-prepared
F. Teng et al. / Journal of Crystal Growth ] (]]]]) ]]]–]]] 7
samples with different amounts of SDBS. Fig. 8 shows the FE-SEMimages of these samples. When no SDBS was added to the system,large nearly spherical particles were obtained (Fig. 8(a)). In thiscase, neither the hierarchically assembled structure (i.e. nano-dendrites) nor the nanorod building blocks were observed,although the final samples seem to be composed of high-purityLiFePO4 crystals. Under an SDBS/Fe molar ratio of 1/1, the optimalsynthesis conditions in our research, a number of well-definedhierarchical structures of nanorods are observed, as exemplifiedby sample S1-2 in Fig. 2. When the molar ratio of SDBS/Fe was
20
(020
)
(201
)
(113
)
(430
)(3
31)
(412
)(222
)(4
21)
(022
)
(112
)
(410
)(2
21)
(121
)(3
11)
(301
)(211
)
(111
)(0
11)
(210
)(1
01)
(200
)
Inte
nsity
/a.u
.
2 Theta/degree
(a)
(b)
(c)
30 40 50 60 70
Fig. 7. XRD patterns of the as-prepared samples corresponding to those in Table 3:
(a) S3-1, (b) S3-3 and (c) S3-4.
Fig. 8. FE-SEM images of the as-prepared samples corresponding to those in Table 3: (a)
Please cite this article as: F. Teng, et al., J. Crystal Growth (2010), do
further increased to 2/1, relatively larger dumbbells of 10 mm inlength are observed, consisting of 3–4 mm long microrods (S3-3 inFig. 8(b)). When the volumetric ratio of SDBS/Fe was furtherincreased to 4/1, coarse bone-shaped microparticles of 10 mm inlength were obtained (S3-4 in Fig. 8(c) and (d)). The blow-up viewof those nanorods in S3-4 is shown in Fig. 8(d). It appears thatthey are assembled by nanorods, similar to the building blocks ofnanodendrites in S1-2. These observations clearly establish thatthe presence of SDBS plays a key role in the formation of nanorodbuilding blocks, as well as the self-assembly process. Additionally,the amount of SDBS is a crucial factor for the control of thehierarchical structures. We further conjecture that SDBS may actas a soft template to direct the growth of nanoparticles intonanorods at the early stage of the process through a preferentialbonding to certain crystal planes. Later on, these nanorods arefurther aligned tightly and assembled end-to-end under theguidance of the soft template to form various forms of hierarchicalstructures (e.g., nanodendrites or bones).
3.4. Formation mechanism of nanodendrites
From the results obtained in this research, it can be concludedthat the EG/W mixture system and SDBS are essential for theformation of novel hierarchical LiFePO4 structures. First of all, thereactivity of all the precursor chemicals can be reduced in thismixture solvent medium. At an optimal EG content (EG/W¼1/1 inour experiment), the differences among the reactivity of theprecursors would be reduced significantly so that their reactivitycan match each other, which will facilitate the formation of thedesired olive-structure crystal instead of the impurity crystals,such as Li3PO4 and Fe4(PO4)3(OH)3. Secondly, the solubilities ofthe precursors in EG/W would be smaller than those in deionized
S3-1, (b) S3-3, (c) S3-4 and (d) the tip of the particles (indicated with arrow in (c)).
Fig. 11. (a) Charge/discharge curves at different discharge rates from 0.1C to 5C
and (b) discharge capacity vs. cycling number at a capacity rate of 0.1C of LiFePO4
nanodendrites between 2.0 and 4.2 V.
F. Teng et al. / Journal of Crystal Growth ] (]]]]) ]]]–]]]8
water, i.e. the precursor chemicals in EG/W could have a higherdegree of supersaturation than they do in deionized water if thesame amounts of precursors are used. Consequently, the EG/Wsystem would favor the nucleation and growth processes ofcrystals. At too high an EG content, the precursors cannot bemixed homogeneously due to the lower solubility of theprecursors. For the process reported herein, the mixture systemturns out to be a promising reaction medium to provide well-controlled crystallization. Moreover, a proper surfactant presentin the system can be used to adjust the size and morphology ofthe particle being produced by binding SDBS molecules onto thenewly formed surfaces during crystal growth [18,28–31]. Theformation mechanism of the hierarchical structures was proposedand described in Fig. 9. The time-dependent experiment resultspresented in Table 1 and Fig. 4 reveal the evolution of thehierarchical nanostructures, while the results at various SDBScontent (Table 3 and Fig. 8) have revealed the critical role playedby this surfactant in the formation of the hierarchical structure. Aswe concluded in the previous section, SDBS, an anionic surfactant,was used to both guide the growth of the crystalline LiFePO4
nanorods and template their self-assembly. During crystalgrowth, the surfactant acted as strong coordinating agents bybinding to some crystal faces, and accordingly inhibited thecrystal growth along the other defined crystal plane. As a result, atthe early stages of the process, SDBS directs crystals to grow alongthe crystal direction whose crystal plane weakly bonds with thesurfactant molecules. As a result, nanorods are formed, as was
Please cite this article as: F. Teng, et al., J. Crystal Growth (2010), do
observed and described in the former parts. In comparison, in theabsence of SDBS, no nanorods building blocks were found in thefinal product (Fig. 8(a)). In addition, it is conjectured that the SDBSmolecules also act as a soft template to assemble the nanorodbuilding blocks into the final hierarchical structures, in which thenanorods are tightly attached together by their ends. We assume
F. Teng et al. / Journal of Crystal Growth ] (]]]]) ]]]–]]] 9
that the end-to-end self-assembly results from the Van der Waalsattraction of hydrophobic interaction of the surfactant moleculesbonded to the end of the nanorods. Since surfactant moleculespreferentially and strongly adsorbed on the nanorod side, thestronger electrostatic force exists on the nanorod side than thaton the end. It seems that there is a balance between electrostaticrepulsion interaction and hydrophobic attraction interaction.Hence the electrostatic repulsion interaction on the side isstronger than that on the end between nanorods, which refrainsthe side-by-side attachment. The long hydrophobic chains of theSDBS molecules boned on the nanorods will be attracted to oneanother through hydrophobic interaction. As a result, thenanorods are attached with each other by their ends to formhierarchical structure. Note that this hypothetic formationmechanism still needs to be confirmed by direct proof. Summa-rily, the reported approach demonstrates obvious advantages,such as one-step synthesis, environmental friendliness, and lowcost. The methodology will also inspire a similar route to prepareother phospholivines, e.g., LiMnPO4. Most importantly, nanoden-drites have been recently identified as one of the most promisingmorphologies for ultrahigh electrochemical/catalytic activity [32].The synthesized LiFePO4 nanodendrites could be a promisingelectrode material for lithium-ion batteries.
3.5. Electrochemical properties
The electrochemical properties of LiFePO4 nanodendrites wereinvestigated and compared with other three typical samples.Fig. 10 shows the charge/discharge profiles of the samples in thecutoff voltage range of 2.0–4.2 at 0.1C (17 mA g�1). A flat plateauat 3.4�3.5 V can be observed obviously during the charge/discharge process, representing the typical electrochemical Li+
insertion/extraction behavior of LiFePO4. This flat plateau isattributed to the two-phase reaction as follows (1) [33]:
LiFePO42(1�x) LiFePO4+xFePO4+xLi+ +xe� (1)
At 0.1C, the discharge capacity of LiFePO4 nanodendrites (S1-2sample) is determined to be 154 mA h g�1 for the first cycle,which is much higher than those (103, 107, 110 mA h g�1) of theS3-1, S3-3 and S3-4 samples. This difference may be explained bytheir different microstructures or textural properties. To furtherconfirm it, the BET areas of the S1-2 and S3-4 samples have beentested with the isothermal sorption of nitrogen. The BET area ofLiFePO4 nanodendrites is determined as 32.6 m2 g�1, much largerthan that (3.4 m2 g�1) of the S3-4 sample. The large surface areaof the active material facilitates its sufficient contact with theelectrolyte, resulting in high Li ion mobility and low charge-transfer resistance. Moreover, the higher BET area of thenanodendrites may also confirm the hierarchical microstructuresconstructed by the nanorods. These small nanorods may facilitatethe intercalation processes of lithium ions, which may haveimproved the charge and discharge kinetics. Martin et al. havereported that the nanofiber morphology mitigates the slowtransport problem of Li ions, because the diffusion distance of Liions within the electrode material is minimized [34]. Therefore,the higher discharge capability of LiFePO4 nanodendrites can bemainly ascribed to the shorter and easier diffusion path of Li ionsin the material with small grain size [33,35]. Fig. 11a gives thecharge/discharge curves of LiFePO4 nanodendrites at differentdischarge rates from 0.1C to 5C. Its initial discharge capacityalmost maintained 154 mA h g�1 at 0.5C. Furthermore, thedischarge capacities at the rates of 1C and 5C decreased to 145and 141 mA h g�1, respectively, corresponding to about 93% and90% capacity retention, compared with that at a rate of 0.1C. It can
Please cite this article as: F. Teng, et al., J. Crystal Growth (2010), do
thus be confirmed that the nanodendrites show good ratecapacity. Fig. 11(b) gives the cycling stability of the nanodendritesat a 0.1C rate. There was almost no capacity fading after 50 cycles,indicating its good cycling stability. The good capacitive behaviorof the hierarchical LiFePO4 nanodendrites demonstrates itspotential application as an excellent cathode material forlithium-ion batteries.
4. Conclusions
LiFePO4 nanodendrites composed of nanorod building blockshave been successfully synthesized in the EG/W system usingSDBS as the surfactant. The results indicate that both reactionmedium (EG/W) and SDBS played an important role in the self-assembly of hierarchical nanostructures. Evidences are shownthat the nanodendrite hierarchical structures are formed throughthe self-assembly of nanorods by their end-to-end attachment.The reported approach is simple and economical, which hassimplified the traditional multi-step methods and will encouragethe future applications of LiFePO4 crystals, especially in lithium-ion batteries.
Acknowledgments
We would like to thank Mr. Nate Kroodsma for Englishimprovement.This work was financially supported by the Michi-gan Tech Faculty Startup Fund, Michigan Tech Research ExcellentFund, and the Korea Research Foundation grant (KRF-2007-412-J04003).
References
[1] A.K. Padhi, K.S. Nanjundaswamy, J.B. Goodenough, J. Electrochem. Soc. 144(1997) 1188.
[2] J.-M. Tarascon, M. Armand, Nature 414 (2001) 359.[3] P.S. Herle, B. Ellis, N. Coombs, L.F. Nazar, Nat. Mater. 3 (2004) 147.[4] B. Kang, G. Ceder, Nature 458 (2009) 190.[5] C. Delacourt, P. Poizot, J.M. Tarascon, C. Masquelier, Nat. Mater. 4 (2005) 254.[6] Q. Wang, S.M. Zakeeruddin, D. Wang, I. Exnar, M. Gratzel, Angew. Chem. Int.
Ed. 45 (2006) 8197.[7] S.Y. Chung, J.T. Bloking, Y.M. Chiang, Nat. Mater. 1 (2002) 123.[8] I. Belharouak, C. Johnson, K. Amine, Electrochem. Commun. 7 (2005) 983.[9] N. Ravet, A. Abouimrane, M. Armand, Nat. Mater. 2 (2003) 702.
[10] Y.S. Hu, Y.G. Guo, R. Dominko, M. Gaberscek, J. Jamnik, J. Maiar, Adv. Mater. 19(2007) 1963.
[11] F. Croce, A.D. Epifanio, J. Hassoun, A. Deptula, J. Olczac, B. Scrosati,Electrochem. Solid-State Lett. 5 (2002) A47.
[12] G.X. Wang, L. Yang, Y. Chen, J.Z. Wang, S. Bewlay, H.K. Liu, Electrochim. Acta50 (2005) 4649.
[13] P.P. Prosini, D. Zane, M. Pasquali, Electrochim. Acta 46 (2001) 3517.[14] C. Delacourt, P. Poizot, M. Morcrette, J.M. Tarascon, C. Masquelier, Chem.
Mater. 16 (2004) 93.[15] H.C. Kang, D.K. Jun, B. Jin, E.M. Jin, K.H. Park, H.B. Gu, K.W. Kim, J. Power
Sources 179 (2008) 340.[16] K.F. Hsu, S.Y. Tsay, B.J. Hwang, J. Mater. Chem. 14 (2004) 2690.[17] S. Yang, Y. Song, P.Y. Zavalij, M.S. Whittingham, Electrochem. Commun. 4
(2002) 239.[18] B. Ellis, W.H. Kan, W.R.M. Makahnouk, L.F. Nazar, J. Mater. Chem. 17 (2007)
3248.[19] Y. Zhang, H. Feng, X. Wu, L. Wang, A. Zhang, T. Xia, H. Dong, M. Liu,
Electrochim. Acta 54 (2009) 3206.[20] K. Konstantinov, S. Bewlay, G.X. Wang, M. Lindsay, J.Z. Wang, Electrochim.
Acta 50 (2004) 421.[21] K. Dokko, K. Shiraishi, K. Kanamura, J. Electrochem. Soc. 152 (2005) A2199.[22] G. Chen, X. Song, T.J. Richardson, Electrochem. Solid-State Lett. 9 (2006) A295.[23] C.R. Sides, F. Croce, V.Y. Young, C.R. Martin, B. Scrosati, Electrochem. Solid
State. Lett. 8 (2005) A484.[24] K. Saravanan, M.V. Reddy, P. Balaya, H. Gong, B.V.R. Chowdari, J.J. Vittal,
J. Mater. Chem. 19 (2009) 605.[25] R. Dominko, M. Bele, J.-M. Goupil, M. Gaberscek, D. Hanzel, I. Arcon, J. Jamnik,
Chem. Mater. 19 (2007) 2960.[26] D. Rangappa, K. Sone, T. Kudo, I. Honma, J. Power Sources, in press,
doi: 10.1016.[27] H. Yang, X.-L. Wu, M.-H. Cao, Y.-G. Guo, J. Phys. Chem. C 113 (2009) 3345.
F. Teng et al. / Journal of Crystal Growth ] (]]]]) ]]]–]]]10
[28] J. Xiang, H. Cao, J.H. Warner, A.A.R. Watt, Cryst. Growth Des. 8 (2008)4583.
[29] G. Zhou, M. Lu, Z. Yang, F. Tian, Y. Zhou, A. Zhang, Cryst. Growth Des. 7 (2007)187.
[30] Y. Huang, W. Wang, H. Liang, H. Xu, Cryst. Growth Des. 9 (2009) 858.[31] G. Leem, S. Sarangi, S. Zhang, I. Rusakova, A. Brazdeikis, D. Litvinov, T.R. Lee,
Cryst. Growth Des. 8 (2009) 32.
Please cite this article as: F. Teng, et al., J. Crystal Growth (2010), do
[32] B. Lim, M. Jiang, P.H.C. Camargo, E.C. Cho, J. Tao, X. Lu, Y. Zhu, Y. Xia, Science324 (2009) 1302.
[33] P. Gibot, M.C. Cabanas, L. Laffont, S. Levasseur, P. Carlach, S. Hamelet,J.M. Tarascon, C. Masquelier, Nat. Mater. 7 (2008) 741.
[34] C.J. Patrissi, C.R. Martin, J. Electrochem. Soc. 146 (1999) 3176.[35] N. Meetong, H. Huang, W.C. Carter, Y.M. Chiang, Electrochem. Solid-State Lett.
