Continuous flame aerosol synthesis of carbon-coated nano-LiFePO₄ for Li-ion batteries

Abstract

Core-shell, nanosized LiFePO₄-carbon particles were made in one step by scalable flame aerosol technology at 7 g/h. Core LiFePO₄ particles were made in an enclosed flame spray pyrolysis (FSP) unit and were coated in-situ downstream by auto thermal carbonization (pyrolysis) of swirl-fed C₂H₂ in an O₂-controlled atmosphere. The formation of acetylene carbon black (ACB) shell was investigated as a function of the process fuel-oxidant equivalence ratio (EQR). The core-shell morphology was obtained at slightly fuel-rich conditions (1.0 < EQR < 1.07) whereas segregated ACB and LiFePO₄ particles were formed at fuel-lean conditions (0.8 < EQR < 1). Post-annealing of core-shell particles in reducing environment (5 vol% H₂ in argon) at 700 °C for up to 4 hours established phase pure, monocrystalline LiFePO₄ with a crystal size of 65 nm and 30 wt% ACB content. Uncoated LiFePO₄ or segregated LiFePO₄-ACB grew to 250 nm at these conditions. Annealing at 800 °C induced carbothermal reduction of LiFePO₄ to Fe₂P by ACB shell consumption that resulted in cavities between carbon shell and core LiFePO₄ and even slight LiFePO₄ crystal growth but better electrochemical performance. The present carbon-coated LiFePO₄ showed superior cycle stability and higher rate capability than the benchmark, commercially available LiFePO₄.

1. Introduction

The olivine LiFePO₄ is attractive for high volume Li-ion battery applications (e. g., in hybrid or solely electric vehicles) (Padhi et al., 1997) as it is made from low cost raw materials, shows high cycle stability, is environmentally benign (Hong et al., 2009). Limitations of LiFePO₄ are its low electric conductivity and slow Li-ion diffusion along the one-dimensional channels of its olivine structure (Morgan et al., 2004). The diffusion limitation can be addressed with smaller particles allowing higher charge- and discharge rates. The conductivity can be increased by Nb doping (Chung & Chiang, 2003), or embedding LiFePO₄ in a conductive matrix (Chen & Dahn, 2002). Carbon black containing materials have typically 10-100 times higher electrical conductivity than doped LiFePO₄ (Delacourt et al., 2006b). As dispersion of the conductive additive is particularly important for nano-sized particles (Patey et al., 2009b), a core-shell structure with active material as core and conductive coating as shell is ideal for assembling conductive nanostructures (Liu et al., 2006). For example, core-shell particles can minimize cell polarization as each particle receives electrons from all possible directions (Wang et al., 2008). Furthermore, the carbon shell can protect the core particle from reactions with the electrolyte as well as during its handling or processing.

Two main strategies to produce carbon-coated battery electrode materials are reported: (1) Admixing of (prefabricated) conductive carbon black during or after synthesis of the active material followed by intimate mixing/milling and high temperature annealing (Shin et al., 2006). The carbon black layer around LiFePO₄ formed by this route is typically rough and discontinuous (Shin et al., 2006). (2) Co-synthesis of active material and carbon black to facilitate excellent dispersion at the nanoscale. This excellent dispersion has been achieved by flame aerosol technology (Spicer et al., 1998) that is able to produce active materials (LiMn₂O₄, Li₄Ti₅O₁₂, LiFe₅O₈) for batteries (Ernst et al., 2007). This technology is quite attractive for its capacity to manufacture nanostructured ceramic materials at large scale, a key feature for reduction of the high cost of battery materials today (Masquelier, 2010). Furthermore, FSP can control product characteristics like size and crystallinity by varying precursor and oxidant composition and flow rate (Strobel & Pratsinis, 2007). Mixed acetylene carbon black (ACB) and LiMn₂O₄ have been made sequentially by separate open flames and the resulting, well-dispersed conductive ACB significantly improved the rate performance of nano-LiMn₂O₄ (Patey et al., 2009a). Core-shell particles are an alternative structure promising even better contact between active material and carbon black.

Here ACB-coated LiFePO₄ particles with core-shell morphology are made continuously, for the first time to our knowledge, by flame aerosol technology, capable to produce, at least, up to 700 g/h of ACB-coated ceramic nanoparticles (Kammler et al., 2001). The influence of O₂-stoichiometry, LiFePO₄-ACB morphology, and annealing on the core-shell morphology and LiFePO₄ crystallinity was investigated. The electrochemical performance (cycle stability & discharge rate capacity) of these core-shell LiFePO₄-ACB particles is presented and compared to that of commercial LiFePO₄ particles dispersed in carbon black.

2. Experimental

2.1 Apparatus

The in-situ ACB coating setup was developed and synthesis parameters were optimized based on flame aerosol synthesis and in-situ coating of particles (Ernst et al., 2008; Patey et al., 2009a; Teleki et al., 2008). Fig. 1 shows a schematic of the experimental setup for synthesis of core-shell LiFePO₄-ACB particles. To prevent air entrainment, the flame reactor is enclosed by quartz glass tubes of 5.2 cm outer diameter and 0.25 cm wall thickness. The combustion environment can therefore be controlled precisely (Heine & Pratsinis, 2005). At 40 and 55 cm above the burner, gaseous C₂H₂ and N₂, respectively, are injected into the gas stream dividing the reactor in 3 zones: (1) An O₂-rich core particle formation zone, (2) an O₂-lean C₂H₂-pyrolysis and carbon-coating zone, and (3) in a quenching and cooling zone.

Fig. 1.

Schematic of the enclosed FSP setup with a particle formation zone (1), an ACB-coating (2), and a quenching (3) zone. By enclosing the unit with quartz tubes, the O₂ stoichiometry in the ACB-coating zone can be controlled precisely. The coated particles are cooled with N₂ at the end of the coating zone to avoid carbon black combustion after contacting ambient air on the way to filtration.

In the lowest zone 1, core LiFePO₄ particles are formed (details in section 2.2). A mixture of O₂ and N₂ (Pan Gas, ≥ 99.95) was fed as sheath gas with a constant flow rate of 19 L/min through 12 circular openings of 0.1 cm diameter surrounding the FSP burner at a radius of 2.2 cm from the nozzle center. The mixing ratio of N₂ and O₂ in this sheath gas was used to control the global equivalence ratio (EQR) in the reactor (Turns, 1996):

$$\mathrm{EQR}=\frac{{\left({\mathrm{n}}_{\text{fuel}}/{\mathrm{n}}_{\text{oxidant}}\right)}_{\text{real}}}{{\left({\mathrm{n}}_{\text{fuel}}/{\mathrm{n}}_{\text{oxidant}}\right)}_{\text{stoichiometric}}}$$ (1)

with n_fuel and n_oxidant being the number of moles of fuel (LiFePO₄ precursor solution, CH₄ and C₂H₂) and oxidant (O₂), respectively. The ratio in the numerator (real) represents the actual moles fed into the system, whereas the ratio in the denominator (stoichiometric) represents the moles needed theoretically for stoichiometric combustion of the above fuel to CO₂ and H₂O. All combustibles (LiFePO₄ precursor solution, CH₄ and C₂H₂) were regarded as fuel and the sum of all supplied O₂ as oxidant.

At the beginning of zone 2 (at 40 cm above the FSP burner), 1 L/min C₂H₂ (Pan Gas, ≥ 99.5 %) was introduced through a stainless steel torus ring. For better gas mixing, a swirl motion is introduced by the radial alignment of the 12 equidistant circular outlets (0.05 cm in diameter) of the ring (Teleki et al., 2009). The aerosol of freshly-made LiFePO₄ particles then passes through the coating section (zone 2) of 15 cm length before it is cooled with N₂ from about 600 to below 300 °C in zone 3 to prevent oxidation of ACB in ambient air. This cooling is achieved by swirl-injection of 20 L/min N₂ through a second torus ring at the entrance of zone 3. A third glass tube, 10 cm long, on top of that torus ring ensured mixing of N₂ and product LiFePO₄-ACB aerosol before its release to ambient air and particle collection on a glass-fiber filter (Whatman GF6, 25.7 cm in diameter) with the aid of a vacuum pump (Busch, Seco SV 1040C).

2.2 Particle Synthesis and Characterization

Core LiFePO₄ nano-particles were made by FSP (Madler et al., 2002). Li-acetylacetonate (Sigma-Aldrich, 97 %), iron (III)-acetylacetonate (Fluka, ≥ 99 %) and tributyl phosphate (Fluka, ≥ 99 %) were mixed stoichiometrically (mol ratio Li/Fe/P = 1/1/1) and were dissolved in an equivolumetic mixture of 2-ethylhexanoic acid (Riedel-de Haën, ≥ 99 %), toluene (Riedel-de Haën), diethylene glycol monobutyl ether (Fluka, purum ≥ 98 %) and ethanol (Alcosuisse, 98 %) resulting in a precursor solution of 0.24 M (total metal concentration). That solution was fed through the FSP nozzle at 3 mL/min and dispersed by 3 L/min of O₂ (Pan Gas, > 99.95 %) whereas 1.6 bar pressure drop was maintained over that nozzle. The spray was ignited by a support premixed flame of 2.5 L/min O₂ and 1.25 L/min CH₄ (Pan Gas, ≥ 99.5 %) through an annulus surrounding the dispersion O₂ outlet (Madler et al., 2002).

Pure LiFePO₄ was made in the same setup as with core-shell particles with EQR = 0.8 and 0.98 but without C₂H₂ flow. Pure ACB was made by spraying the above solvent solution without the Li, Fe, and P precursors. Co-synthesized ACB and LiFePO₄ was made with two flames as described earlier (Patey et al., 2009a) using the above precursor solution.

Powder annealing was carried out in a thermo balance (Mettler-Toledo, TGA/SDTA851e) using 900 μL alumina crucibles 2/3 filled with slightly compressed (~20 kPa) sample material. Heating and cooling rates were both set to 10 °C/min. The annealing time is defined as the isothermal time of the run. Annealing was carried out by supplying 200 mL/min of 5 vol% H₂ in argon (Hydrargon 5, Pan Gas). Low-temperature oxidation of in-situ ACB-coated LiFePO₄ was made in a muffle oven (Barnstead International, Thermolyne 48000) at 420 °C in open ceramic crucibles and still air atmosphere. Heating and cooling rates were set to 10 °C/min and the oxidation time is defined as the isothermal time of the run.

Powders were characterized by N₂ adsorption (specific surface area, SSA), X-ray diffraction (XRD) and transmission electron microscopy (TEM) For the calculation of the Brunauer–Emmett–Teller equivalent diameter (d_BET) of pure LiFePO₄ particles a LiFePO₄ density of 3.577 g/cm³ was used (Zane et al., 2004). The ACB content (X_ACB) was determined thermogravimetrically (Mettler Toledo TGA/SDTA851e) by heating the as-prepared sample in a 70 μL alumina crucibles in O₂ to 800 °C with 10 °C/min. The X_ACB was defined as the relative weight loss between 120 and 800 °C, as below 120 °C typically physically bound water is desorbed (Mueller et al., 2003). Particle size distributions (PSD) and their average diameter and geometric standard deviation were obtained by manually counting TEM particle images (Hinds, 1999).

2.3 Electrochemical Characterization

Dry powder of annealed core-shell LiFePO₄-ACB made at EQR = 1.02 was used as active material. This material showed the most promising coating uniformity and thickness by TEM and was therefore used for all electrochemical evaluations. For slightly rougher coatings (e.g. EQR 1.07) one expects similar annealing as the also continuous but rougher ACB shell hinders particle sintering. Regarding electrochemical properties, rougher coatings are undesired, as they decrease the specific charge of the electrode by adding inactive mass and might even represent an additional Li⁺ diffusion resistance by the protruding asperities (contact resistance).

Commercial LiFePO₄ (Sigma-Aldrich, LiFePO₄, SSA = 14.5 m²/g, battery grade) with d_BET = 116 nm and d_XRD = 150 nm served as benchmark material and was processed into composite electrodes in the exact same way as the core-shell LiFePO₄-ACB. The commercial material serves here as a common reference since the electrode preparation technique has great impact on the electrochemical performance (Marks et al., 2011). Carbon black (CB; Super P, TIMCAL SA, SSA = 62 m²/g) was used as an electrically conductive additive. Polyvinylidene fluoride (PVDF SOLEF 1015, Solvay) dissolved in N-methylpyrrolidinone (NMP, Fluka) was used as a binder. The mass-ratio (after drying) of AM/CB/PVDF in the electrodes was 83/8.5/8.5. To prepare the suspensions, all electrode constituents were mixed in NMP with an ultrasonic stirrer (Hielscher UP 200H) for 2 min at 200 W and 24 kHz. To prepare test electrodes, all suspensions were doctor-bladed at a “wet” thickness of 250 μm onto a carbon-coated aluminum current collector (GAIA, Nordhausen) and dried at 110 °C for 10 min at atmospheric pressure and under vacuum overnight. Afterwards, disk electrodes with a diameter of 1.3 cm were punched out and dried under vacuum at 120 °C overnight. Coin-type test cells were assembled in an argon-filled glove box with contents of less than 1 ppm of O₂, N₂, and H₂O. Lithium metal (99.9 %, Aldrich) served as both reference and counter electrode (anode). A fiberglass separator (1 mm thick) was soaked with 500 μl of electrolyte (1 M LiPF₆ in ethylene carbonate (EC) / dimethyl carbonate (DMC), 1:1 by mass, Ferro). This electrolyte was mixed with a detergent (Triton X-209, 2 wt%, Sigma Aldrich) prior to soaking the separator.

All electrochemical measurements (cycling and power protocol) were performed at 25 (±0.1) °C and a constant specific current corresponding to 0.5C - 16C (1C = 170 mA/g of active material, LiFePO₄). For the rate capability experiments, the electrodes were cycled galvanostatically between 2.0 and 4.5 V vs. Li/Li⁺ for varying specific currents proportionally to the LiFePO₄ content. These experiments were performed after 50 standard cycles at 2C rate. To promote complete charge/discharge at the respective potential limits, a potentiostatic step was included until the specific current decreased to 34 mA/g. All results are the average of three tests. The capacity density was obtained from the specific charge (mAh/g active material, LiFePO₄) and the composite electrode volume, calculated from electrode projected area and thickness, respectively. Capacity fading is defined as the loss of specific discharge capacity per cycle, relative to the initial discharge capacity.

3. Results and Discussion

3.1 LiFePO₄-ACB Morphology

Fig. 2a shows a TEM picture of as-prepared, pure LiFePO₄ particles. Fig. 2b,c,d show as-prepared powders made at EQR of 0.89, 1.02, and 1.07, respectively, by changing the O₂-content in the sheath gas from 35 to 19 vol% at constant 1 L/min C₂H₂. It should be noted that other operation conditions, e.g. C₂H₂ injection height, injection rate, coating tube length also affected product particle characteristics (e.g. Teleki et al., 2008; 2009). The EQR, however, showed the strongest impact on ACB morphology (core-shell vs. segregated particles) and was therefore discussed preferentially here. The powders made at the above EQR contain 53, 18, and 10 wt% carbon black, respectively. As the particles in Fig. 2a do not contain visual carbon residues, the carbon black source in Fig 2b,c,d is the C₂H₂. Fig. 2b shows segregated single LiFePO₄ particles and agglomerated ACB containing primary particles of approximately 10 nm diameter each despite excess of O₂ in the overall system during synthesis. No coating is visible on the surface of LiFePO₄ in Fig. 2a,b. In Fig. 2c, the core LiFePO₄ particles are coated with a continuous and uniform 5 nm thick ACB layer. At higher EQR of 1.07 (Fig. 2d), the ACB layer thickness is slightly rougher with a tendency to form necks between the ACB-coated particles. The change from segregated to shell ACB morphology can be attributed to the different mode of carbon formation under fuel-rich and -lean conditions:

Fig. 2.

TEM images of as-prepared LiFePO₄ and acetylene carbon black (ACB) nanoparticles: a) pure LiFePO₄, b) segregated ACB and LiFePO₄ made at EQR = 0.89. Core-shell LiFePO₄-ACB made at EQR = c) 1.02 and d) 1.07.

No ACB coating was seen for C₂H₂-lean conditions (EQR < 1) as excess O₂ is available. Iron oxides, however, are known to be good oxidation catalysts thus also LiFePO₄ may rather oxidize C₂H₂ on its surface than allowing deposition of ACB. A similar effect was observed during FSP synthesis of carbon embedded Pt clusters where the presence of Pt reduced the carbon black yield by 80 % through catalytic burnoff (Ernst et al., 2008). For example, Fe₂O₃ utilizes surface O₂ (available at EQR < 1) for the catalytic carbon black oxidation at 400 °C (Mul et al., 1998) as Fe₂O₃ can serve as an “O₂-pump” (Reichert et al., 2010). Iron oxides are used also in diesel particle filters to oxidize soot particles (Wenger et al., 2008). Consequently, LiFePO₄ particles do not behave like core SiO₂ (Spicer et al., 1998) or TiO₂ particles (Kammler & Pratsinis, 2003) that act as seeds facilitating surface growth of carbon black in atmospheric diffusion flames.

Additionally, at EQR < 1 the flame temperature is high by the exothermic oxidation of C₂H₂ favoring even more the catalytic activity of LiFePO₄ for surface oxidation of ACB. As a result, mostly segregated ACB is formed at locally fuel-rich environments, preferentially without LiFePO₄ contact.

For C₂H₂-rich conditions (EQR > 1), sticky, tar-like hydrocarbons were formed as indicated by a brownish precipitation on the backside of the glass fiber filters. Such products of incomplete combustion can easily bind on the LiFePO₄ particle surface and therefore enhance formation of ACB coating shells (Fig. 2c,d). Furthermore, the oxidation activity of iron, in LiFePO₄, is too low at fuel-rich conditions so hydrocarbons are available for ACB surface formation.

Fig. 3 shows the ACB content (X_ACB, circles) of the as-prepared powders as a function of EQR. All production parameters were kept constant (except O₂ content in the sheath gas). For increasing EQR, the X_ACB increases, reaching a maximum at EQR = 0.9 and decreases thereafter. For EQR approaching 0.9, ACB is increasingly formed by C₂H₂-pyrolysis as less O₂ is available for C₂H₂ combustion (consistent with the decreasing brightness of the flame). For higher EQR, X_ACB decreases as less C₂H₂ is combusted and pyrolyzed hence reducing the conversion of C₂H₂ to ACB. Core-shell coatings were made in red to bluish C₂H₂ flames (1.0 ≤ EQR ≤ 1.07) spread over the whole length of zone 2. For EQR > 1.1 the C₂H₂ coating flame extinguished, thus no ACB coating was formed anymore and the experiments were stopped for safety reasons.

Fig. 3.

ACB content (○) and SSA of as-prepared composite LiFePO₄-ACB (◇), pure ACB (◻), and pure LiFePO₄ (⟡) as well as the SSA of LiFePO₄-ACB particles after 4 (▽) and 8 hours (△) oxidation in air at 420 °C as a function of the global EQR in the enclosed setup. Core-shell morphology is detected for (1.02 ≤ EQR ≤ 1.07) where the SSA of the composite LiFePO₄-ACB material approaches that of pure LiFePO₄ particles.

Fig. 3 shows also the SSA of as-prepared pure ACB (squares), LiFePO₄ (stars) and LiFePO₄-ACB (diamonds) particles. The SSA of pure LiFePO₄ is constant, at about 30 m²/g, over the entire EQR range as here LiFePO₄ particles (≈57 nm in diameter) are formed in zone 1 before being exposed to C₂H₂. Pure ACB always had much higher SSA than pure LiFePO₄ in particular at fuel-lean conditions (≈ 300 m²/g) with a characteristic drop (≈ 150 m²/g) at fuel-rich EQR (Ernst et al., 2008).

The SSA of the as-prepared LiFePO₄-ACB composite particles with mixed (Fig. 2b) or core-shell morphology (Fig. 2c,d), closely follows the evolution of X_ACB indicating that ACB determines the SSA variation. To elucidate this further, Fig. 3 shows also the SSA as function of EQR of LiFePO₄-ACB, post-oxidized for 4 (inverse triangles) and 8 hours (triangles) in air at 420 °C. The SSA of composite particles gradually converges to that of pure LiFePO₄ (stars) at all EQR (Fig. 3). After this 4-hour low temperature oxidation, powders made at 0.8 ≤ EQR ≤ 0.93 still contain residual carbon, as measured by Raman analysis (not shown). Increasing this oxidation time to 8 hours results in carbon free, amorphous LiFePO₄ similar to as-prepared pure LiFePO₄. The SSA of these low-temperature oxidized particles (≈ 30 m²/g) is the same as for pure LiFePO₄ independently of EQR and initial carbon content in the composite. This is similar to flame-made ACB-coated SiO₂ (Spicer et al., 1998) annealed in air to oxidize away their carbon layer and reveal the finer nanostructure of core SiO₂. Clearly ACB formation does not influence the core-particle SSA as LiFePO₄ formation is finished before ACB is coated as shell or formed as segregated particles.

3.2 Crystallinity & Annealing of Core-shell LiFePO₄-ACB Particles

Fig. 4 shows XRD spectra of as-prepared ACB-coated LiFePO₄ made at EQR = 1.02 (Fig. 2c) and after annealing in Ar/H₂ at 700 & 800 °C for 30 - 240 minutes. As-prepared particles are amorphous by XRD though tiny spikes attributed to nanocrystalline LiFePO₄ could be traced (e.g. at 25° & 35°). After 30 minutes at 700 °C distinctly crystalline LiFePO₄ is formed. Extending the annealing time to 60 and 240 minutes improved the crystallinity, as seen by the reduction of the amorphous hump at 20° < 2θ < 40°. At 800 °C and 240 minutes the amorphous part (hump) disappeared totally, but Fe₂P crystals were formed as ACB can reduce FePO₄ to FeP₂ (Herle et al., 2004). The presence of Fe₂P, however, is not unwelcome here, in contrast to other crystalline phases, as it increases the electronic conductivity of the composite by several orders of magnitude by forming a grain-boundary nano-network of Fe₂P and carbon with LiFePO₄ (Herle et al., 2004).

Fig. 4.

As-prepared and annealed (Ar/H₂ atmosphere) XRD spectra of core-shell LiFePO₄-ACB (EQR = 1.02 ). Pure LiFePO₄ phase was obtained at 700 °C, whereas up to 15 wt% Fe₂P was formed at 800 °C.

Fig. 5 shows the morphology of annealed ACB-coated LiFePO₄ (EQR = 1.02). At 700 °C the ACB-shell remains mostly intact and attached to the core LiFePO₄ particles, however, cavities start to form (arrows in Fig. 5a,b). At 800 °C, the core is progressively thinning out: cavities are formed already after 60 minutes (arrows in Fig. 5c) whereas after 240 minutes empty ACB shells and fragments are present (Fig. 5d). Most likely these cavities result from crystallization and subsequently from reduction of LiFePO₄ to Fe₂P and possibly to Li₃PO₄ or P (Yang et al., 2001). However, phosphorus (m/z = 31) was not detected by mass spectrometry of the effluent gas during TGA (not shown) but Fe₂P formation is proven by post XRD analysis (Fig. 4). Similarly cavities between active material and CB were observed for CB-coated LiMnPO₄ after annealing at 500 °C (Doan et al., 2010).

Fig. 5.

TEM images of core-shell LiFePO₄-ACB (EQR = 1.02) after annealing in Ar/H₂ at 700 °C for a) 60 and b) 240 min, as well as at 800 °C for c) 60 and d) 240 min. Morphology and size are generally preserved during annealing at 700 °C but at 800 °C the core LiFePO₄ can disintegrate and may even breaks the ACB-shell. These particles may grow and form larger LiFePO₄ crystals (according to XRD) leaving behind empty ACB shells.

Fig. 6 shows the crystal size of FSP-made pure LiFePO₄ (squares), ACB-coated (EQR = 1.02, triangles), and mixed LiFePO₄ (circles) particles as a function of annealing time at 700 °C and ACB-coated LiFePO₄ at 800 °C (EQR = 1.02, inverse triangles). Pure LiFePO₄ particles (circles) shows substantial crystal growth from 80 - 250 nm as a function of annealing time at 700 °C. The mixed ACB-LiFePO₄ particles grow from 140 to 180 nm indicating some hindering of crystal growth by the presence to ACB. The crystal size of ACB-coated LiFePO₄, however, increases linearly during the first 120 minutes to 65 nm that is close to the BET-size of as-prepared, amorphous, pure LiFePO₄ of 57 nm (Fig. 3) indicating formation of monocrystalline particles and consistent with the TEM images (Fig. 2 and 5). After 120 min the crystal size is constant for ACB-coated LiFePO₄ at 700 °C. At 800 °C, crystal growth is initially similar to 700 °C but is increasing beyond 120 minutes. Perhaps carbothermal LiFePO₄ decomposition to Fe₂P (and possibly Li₃PO₄) steadily consumes the nanothin ACB-shells from the inside (Kadoma et al., 2010). Eventually, as the carbon shell vanishes, core LiFePO₄ particles can meet and sinter to bigger crystals (according to XRD), leaving behind empty ACB-shells (Fig. 5d). Even though some coating-shells break open, crystal growth during annealing of carbon-coated LiFePO₄ was effectively prevented compared to pure LiFePO₄ (squares) or co-synthesized LiFePO₄ with ACB (circles).

Fig. 6.

LiFePO₄ crystal size evolution during annealing at 700 °C in Ar/H₂ for particles with pure LiFePO₄ (◻), co-synthesized LiFePO₄-ACB (○), core-shell LiFePO₄-ACB (△), and core-shell LiFePO₄-ACB at 800 °C (▽). The crystal size of core-shell LiFePO₄-ACB (EQR = 1.02) was preserved at 700 °C whereas at 800 °C the crystal size increased. Pure and co-synthesized LiFePO₄-ACB (homogeneous mixture of amorphous LiFePO₄ and ACB) showed substantial crystal growth during annealing at 700 °C.

Fig. 7 shows TEM images and the corresponding size distributions of in-situ coated LiFePO₄/ACB made at EQR = 1.02 in a) as-prepared and b) after annealing at 800 °C for 240 min. The as-prepared sample shows perfectly spherical particles with a count median diameter of 58 nm and a geometric standard deviation of 1.44. The latter nicely corresponds to that of the self-preserving particle size distribution for coagulation of spherical particles by Brownian motion in the continuum regime (Xiong & Pratsinis, 1991). The annealed sample, however, shows some regions with non-spherical, rather large particles most probably originating from sintering of primary particles after shell-breakup as shown in Fig. 5d. Nevertheless, the spherical shape of the as-prepared particles is mostly preserved during annealing. This favors the stacking and processing properties of the material (Liu et al., 2006). The size distribution of the annealed sample is also nearly lognormal with a slightly broader geometric standard deviation of 1.50 than the as-prepared sample. The volume based median diameter (88 nm) of the annealed sample, derived from the count median diameter (Hatch & Choate, 1929), compares well with the LiFePO₄ crystal size (80 nm) by XRD (Fig. 6) underlining the formation of mostly monocrystalline LiFePO₄ particles despite the appearance of the Fe₂P phase.

Fig. 7.

Low magnification TEM images of a) as prepared b) and after annealing at 800 °C for 240 min core-shell LiFePO₄-ACB (EQR = 1.02) with the corresponding counted particle size distributions.

3.3 Electrochemical Performance

Fig. 8 shows representative electrochemical evaluation (cycling at 2C rate) of core-shell ACB-coated LiFePO₄ (EQR = 1.02) annealed for 240 min at 700 °C and 800 °C in comparison to commercial LiFePO₄ used as a benchmark. Commercial LiFePO₄ shows capacity fading in the order of 0.2 % for the first 100 cycles decreasing to 0.08 % for the following cycles consistent with the literature (Bakenov et al., 2007). At the given experimental conditions the powder annealed at 800 °C shows higher practical specific discharge capacity than the 700 °C annealed and commercial powders. Perhaps the formation of Fe₂P in the 800 °C annealed powder could have enhanced the electrical conductivity of the active material (Song et al., 2008) increasing the practical charge capacity of LiFePO₄-ACB composite (please consider that the active mass was only corrected for the ACB content and not for the impurity phases). Both, 700 and 800 °C annealed core-shell LiFePO₄-ACB powders show remarkable improvement in cycle stability with only 0.05 % and 0,01 % capacity fading per cycle, respectively. This could be attributed to lower polarization of the active material in the well-dispersed ACB matrix (Liu et al., 2008).

Fig. 8.

Galvanostatic discharge capacity per unit mass (specific charge) of core-shell LiFePO₄-ACB annealed at 700 (△) and 800 °C (▽) as well as commercial LiFePO₄ particles (◇) used for preparation of composite electrodes. Cycling was performed at 2.0–4.5 V (vs. Li/Li⁺) at 2C (1C =170 mA/g) and 25 °C.

Fig. 9 shows the rate capability of the electrodes containing annealed core-shell LiFePO₄-ACB and commercial LiFePO₄ particles. Here the galvanostatic discharge capacity is shown for various C-Rates (0.5-16C) as an average over 10 cycles, starting from the 50^th cycle under standard cycling. The in-situ ACB coated LiFePO₄ shows high specific charge of 163 mAh/g for low C-rates (0.5C) corresponding to 96% of the theoretical charge capacity of LiFePO₄ and the resulting capacity density is about 112 ± 4 mAh/cm³, well in line with commercial LiFePO₄-based batteries. The improved conductivity and connectivity of the core-shell LiFePO₄-ACB compared to benchmark LiFePO₄ became significant at higher C-Rates whereas the 800 °C annealed particles performed best. A certain “size effect” due to smaller crystal sizes of core-shell LiFePO₄-ACB compared to commercial LiFePO₄ cannot be excluded (Delacourt et al., 2006a). Nevertheless, core-shell LiFePO₄-ACB annealed at 800 °C (d_XRD = 80 nm) shows better performance in cycle stability and rate capability than 700 °C (d_XRD = 61 nm) annealed, despite 30% larger LiFePO₄ crystal size. In literature, however, best results were obtained for moderate annealing temperatures of 550 °C (Yamada et al., 2001) or 700 °C (Mi et al., 2005) as above these the crystals were sintering decreasing the practical charge/discharge capacity of LiFePO₄.

Fig. 9.

Galvanostatic discharge capacity per unit mass (specific charge) at various C-rates for core-shell LiFePO₄-ACB annealed at 700 (△) and 800 °C (▽) as well as commercial LiFePO₄ particles (◇) used for the preparation of composite electrodes. Electrochemical measurements were performed at 2.0– 4.5 V (vs. Li/Li⁺) at 25 °C. The constant current (galvanostatic) part of the total charge capacity at each C-rate is shown as the average of 10 cycles, after about 50 cycles at standard 2C rate. The indicated error bar at 16C represents the largest variation over all displayed C-rates.

Here, the core-shell morphology hindered LiFePO₄ sintering enabling high temperature annealing with limited sintering and crystal growth. Better crystallinity in conjunction with small particles for high C-rates could therefore be made. The improved crystallinity in high temperature annealed particles reduces blockages in the one-dimensional Li-ion diffusion channels (Yang et al., 2001) and the presence of Fe₂P increases the electrical conductivity of LiFePO₄ (Song et al., 2008). The ACB shell prevented particle sintering completely up to 700 °C and significantly at 800 °C so particles with crystal sizes well below 100 nm were made showing that the present technology has potential for synthesis of cathode material contributing to durable high performance batteries.

4. Conclusions

Nano-scale LiFePO₄ particles that are thinly (≈ 5 nm) coated with acetylene carbon black (ACB) were continuously produced (7 g/h) by a flame aerosol process. Core-shell morphologies were made at slightly fuel-rich conditions (1.02 ≤ EQR ≤ 1.07) while in excess O₂, LiFePO₄ efficiently catalyzed ACB oxidation on its surface, preventing its coating and forming segregated ACB. The ACB-shell did not influence the LiFePO₄ core particle size as LiFePO₄ formation is finished before synthesis and deposition of the ACB shell. ACB-coated particles were annealed at 700 °C in reducing Ar/H₂ resulting in monocrystalline LiFePO₄ particles with d_XRD = 65 nm. Further particle growth was inhibited by the presence of the ACB shell on their surface. Annealing at 800 °C in Ar/H₂ for 4 hours yielded up to 15 wt% Fe₂P by carbothermal reduction of LiFePO₄ through its ACB-shell that facilitated limited LiFePO₄ crystal growth and cavities and even hollow carbon shells. Core-shell LiFePO₄-ACB particles showed high reversible practical charge capacity of about 160 mAh/g, capacity density of about 112 mAh/cm³ and excellent cycling stability with less than 3 % capacity loss after 200 cycles at 2C.