Thermoanalytical, Optical, and Magnetic Investigations on Nanocrystalline Li0.5Fe2.5O4 and Resulting Ceramics Prepared by a Starch-Based Softchemistry Synthesis

Nanocrystalline Li0.5Fe2.5O4 was prepared by a starch-based soft-chemistry synthesis. Calcining of the (LiFe)-gel between 350 and 1000 °C results in Li0.5Fe2.5O4 powders with crystallite sizes from 13 to 141 nm and specific surface areas between 35 and 7.1 m2 g-1. XRD investigations reveal the formation of ordered Li0.5Fe2.5O4. Sintering between 1050 and 1250 °C leads to ceramics with relative densities of 67–95% consisting of grains between 0.3 and 54 μm. As the sintering temperature increases a rising weight loss of the ceramic samples was observed due to the loss of Li2O. Temperature-dependent magnetic measurements indicate a superparamagnetic behaviour for the nano-sized samples. Field-dependent measurements at 3 K of ceramics sintered between 1050 and 1200 °C show increasing saturation magnetization values (Ms) of 70.0–73.0 emu g-1 most likely due to the formation of lithium vacancies and a decrease of the inversion parameter. The magnetization drops down to 67.7 emu g-1 after sintering at 1250 °C caused by the formation of hematite. Diffuse reflectance spectra reveal an indirect allowed band gap decreasing from 1.93 to 1.60 eV depending on thermal treatment. DSC measurements of the order ⇆ disorder phase transition on nano-sized powders and bulk ceramics exhibit transition temperatures between 734 and 755 °C and enthalpy changes (ΔtrsH) ranging from 5.0 to 13.5 J g-1. The linear thermal expansion coefficient was found to be 11.4*10-6 K-1.


Introduction
Li 0.5 Fe 2.5 O 4 (LiFe 5 O 8 ) has interesting technological applications such as microwave, optical isolator, and memory devices because of its high saturation magnetization and Curie temperature [1,2,3]. Lithium ferrite can be also used as electrode material in lithium ion batteries [4,5]. In addition, Li 0.5 Fe 2.5 O 4 act as catalyst for the synthesis of biodiesel and amino ketones as well as photocatalyst for the decomposition of organic compounds and for water splitting [69]. Furthermore, Rezlescu et al. [10] reported on lithium ferrite for gas sensing applications. Moreover, a magnetoelectric effect was found in Li 0.5 Fe 2.5 O 4 as well as in BaTiO 3 Li 0.5 Fe 2.5 O 4 composites [11,12].
Li 0.5 Fe 2.5 O 4 crystallizes in the inverse spinel structure and occurs in an ordered form (-form, SG: P4 3 32) and a disordered one (-form, SG: Fd3 ̅ m) [13]. The reversible ordered  disordered phase transition in bulk material take place at about 750 °C and a Curie temperature of about 630 °C was found [14]. The disordered form (-Li 0.5 Fe 2.5 O 4 ) can be obtained by rapid quenching of samples from high temperatures to room temperature, whereas the ordered spinel phase forms upon slow cooling. At high temperatures, Li 0.5 Fe 2.5 O 4 loses lithium and oxygen which influences the magnetic and electrical properties [15,16]. Polycrystalline Li 0.5 Fe 2.5 O 4 samples are commonly synthesized by the conventional mixedoxide method which requires high calcining temperatures leading to large particles and high sintering temperatures [17]. Whereas, soft-chemistry syntheses require low reaction temperatures to synthesize nano-sized samples. Various soft-chemical syntheses have been reported, such as hydrothermal [18], solvothermal [19], sol-gel [20,21], combustion [22,23] and precursor routes [24]. Closer inspection shows, that various synthesis routes lead to Li 0.5 Fe 2.5 O 4 with traces of hematite [2528]. In order to avoid the formation of iron-rich secondary phases an excess of lithium was used [2931]. The formation of -Li 0.5 Fe 2.5 O 4 at low temperatures was reported for some soft-chemical syntheses [18,32,33].
The aim of this paper is to describe a facile synthesis route using starch to prepare nanocrystalline Li 0.5 Fe 2.5 O 4 powders and ceramic bodies obtained from these powders. Phase evolution during calcination and sintering were monitored by XRD. Furthermore, magnetic measurements between 3 and 300 K were carried out both on calcined powders and ceramic Electronic copy available at: https://ssrn.com/abstract=3619256 bodies. Moreover, we determined the order  disorder phase transition temperature and enthalpy change depending on particle size as well as the linear thermal expansion coefficient of Li 0.5 Fe 2.5 O 4 .
After addition of 0.0123 mol soluble starch (M = 342.30 g mol 1 , Sigma-Aldrich, ACS reagent) the turbid solution was stirred at room temperature until it turned to a highly viscous red gel. This (LiFe)-gel was calcined for 2 h in static air at various temperatures (heating-/cooling rate 5 K min 1 ) leading to Li 0.5 Fe 2.5 O 4 nano powders. To obtain ceramic bodies, the (LiFe)-gel was calcined at 350 °C for 2 h. Then, the resulting powder was mixed with 10 wt% of a saturated aqueous polyvinyl alcohol (PVA) solution as a pressing aid and uniaxially pressed at about 85 MPa into pellets (green density 1.5 g cm 3 ). These pellets were placed on a ZrO 2 fibre mat and sintered to ceramic bodies.

Characterization
X-ray powder diffraction patterns were recorded at room temperature on a Bruker D8-Advance diffractometer, equipped with a one-dimensional silicon strip detector (LynxEye) using Cu-K  radiation and a counting time of 1 s per data point. Crystallite size and the strain parameter were calculated from XRD line broadening (integral peak breadth) using the Scherrer and Wilson equations (software suite WinXPOW [34]). Dilatometric measurements were carried out in flowing synthetic air (50 ml min 1 ) with a rate of 5 K min 1 and a contact force of 0.2 N in a Netzsch TMA 402F3 dilatometer. Simultaneous thermogravimetric (TG) and differential thermoanalytic (DSC) investigations in flowing synthetic air (50 ml min 1 ) were performed using a Netzsch STA 449F5 system. To study the phase transitions in Li 0.5 Fe 2.5 O 4 , the ceramic bodies were crushed to powders and the DSC curves were performed with a rate of 20 K min 1 . The specific surface area (BET) was determined using nitrogen five-point gasphysisorption (Nova touch 2LX, Quantachrome Corporation). The equivalent BET particle diameter was calculated assuming a spherical or cubic particle shape. Scanning electron microscope images were recorded with a Phenom ProX SEM in the backscattered electron mode (BSE). Diffuse reflectance spectra were recorded at room temperature using a Perkin Elmer UVVis spectrometer Lambda 19 with BaSO 4 as white standard. Magnetic Electronic copy available at: https://ssrn.com/abstract=3619256 measurements were carried out using a Quantum Design PPMS9. Hysteresis loops were taken with magnetic field cycling between 90 and + 90 kOe. In addition, the temperature dependent magnetizations were measured between 3300 K using field-cooled (FC) and zerofield cooled (ZFC) conditions. The samples were enclosed in gel capsules whose very small contribution to the measured magnetic moment was subtracted before data evaluation.

Synthesis and powder characterization
The formation of Li 0.5 Fe 2.5 O 4 was examined by thermal decomposition of the (LiFe)-gel in a muffle furnace in static air (heating rate 5 K min 1 , soaking time 2 h). The prepared red (LiFe)-gel is X-ray amorphous (

Sintering behaviour and microstructure of ceramic bodies
Prior to investigation of the sintering behaviour, the (LiFe)-gel was calcined at 350 °C for 2 h.
As mentioned above, this calcination process leads to a light-brown nanocrystalline Li 0.5 Fe 2.5 O 4 powder with a volume-weighted average crystallite size of 13(1) nm. The specific surface area of that powder was determined as 35(3) m 2 g 1 corresponding to a calculated equivalent particle size of 36(3) nm. The difference between the crystallite size and the particle size from BET data can be explained by an agglomeration leading to surface areas unavailable for nitrogen adsorption.
This nanocrystalline powder was pressed to pellets and isothermal sintered for 1 h in static air at different temperatures (heating-/ cooling rate: 5 K min 1 ). The final bulk densities (Fig. 3) of the black-brown ceramic bodies were calculated from their weight and geometric dimensions and related to the single crystal density of 4.72 g cm 3 [13]. Sintering at 1000 and 1050 °C results in a poor densification with relative densities of 67(1) and 79(1) %, respectively. Firing at 1100 °C leads to bodies with 90(1) % relative density which increases to 93(1) and 95(1) % at 1150 and 1250 °C, respectively. SEM images of ceramics bodies are shown in Fig. 4. Ceramics sintered at 1050 °C and 1100 °C show irregular grains with a 4a,b). After firing at 1150 °C the grain sizes range between 1.5 and 36 µm and the average grain size is 6.6(5) µm as determined by the lineal intercept method [35]. The grains grow to 2.545 µm ( li = 11(1) µm) and 2.554 µm ( li = 13(1) µm) at 1200 and 1250 °C, respectively (Fig. 4c,d).  As reported elsewhere, heating of Li 0.5 Fe 2.5 O 4 leads to a sublimation of Li 2 O [36,37]. To estimate the weight loss during thermal treatment, several powder compacts were fast preheated to 800 °C to burn out the organic binder (PVA) and thermogravimetric measurements were carried out on these compacts. Fig. 5 shows a non-isothermal thermogravimetric measurement in flowing air up to 1250 °C (rate 5 K min 1 ). A very slight weight loss starts at 750 °C, whereas a strong increase of the weight loss rate was observed above about 1130 °C. After one hour sintering at 1050, 1100, 1150, 1200, 1250 °C, the lithium loss was calculated as 4, 6, 8, 12, and 15 mol% assuming the total weight loss is caused by the loss of Li 2 O (inset in Fig. 5). A possible partial reduction of Fe 3+ to Fe 2+ was neglected because sintering was carried out in air with low cooling and heating rates [36,37].  Electronic copy available at: https://ssrn.com/abstract=3619256  Additionally, higher calcining temperatures lead to an improved crystallinity and thus to a reduction of defects [44,45]. The root-mean-square strain parameter, calculated from the XRD line broadening, reflects the number of crystal lattice defects and decreases from 5(1)·10 3 to 0.3(1)·10 3 with increasing calcining temperature to 1000 °C.   Fig. 9. The ceramics show ferrimagnetic behaviour. The saturation magnetization increases slightly from 70.0(1) to 73.0(1) emu g 1 with rising sintering temperature up to 1200 °C (inset in Fig. 9). Increasing M s values with rising heat treatment were already reported in literature [37,46,47]. Pointon and Saull [15] and Ridgley et al. [37] assumed that an increasing of the saturation magnetization above the value for an ideal invers spinel structure (M s = 2.5 µ B f.u. 1 [52], detected by XRD (see Fig. 6), as a result of the considerable loss of lithium. DSC measurements up to 900 °C reveal both the transition from the ordered to the disordered structure (~ 750 °C) and the ferrimagnetic  paramagnetic (~ 630 °C) transition ( Fig. 10) [5355]. The ordered  disordered phase transition temperature (T trs ) was determined from the onset of the DSC signal and was found to be T trs = 755 (1) (1) °C, respectively. The corresponding enthalpy changes were calculated as 5.0(6), 6.4(6), disordered phase transition show that a reduction of the particle size leads to a significant smaller enthalpy change and to lower transition temperatures mainly due to a surface effect in small particles [58,59]. The thermoanalytical behaviour of Li 0.5 Fe 2.5 O 4 powders calcined at 800 °C and higher (crystallite size >> 20 nm) are close to the bulk specimens.

Magnetic, thermoanalytic, and optical investigations
The magnetic transition (ferrimagnetic  paramagnetic) is clearly seen in DSC curves of the sintered ceramics as a weak broad peak (inset in Fig. 10). The Curie temperature (onset temperature) for all sintered ceramics could be estimated as 625 (5)    Calculations up to 700 °C (below the phase transition) reveal a linear thermal expansion coefficient of  dil = 11.4(7)10 6 K 1 in good agreement with the value found by Kato [56]. Diffuse reflectance spectra of calcined and sintered Li 0.5 Fe 2.5 O 4 samples were recorded to determine the optical band gap using the KubelkaMunk theory [61,62] in which the optical band gap can be expressed by Eq. 2 [63]: assuming direct allowed as well as indirect allowed transition mechanisms [8,9,27,64]. To verify the type of transition, the exponent n can be estimated by linearization of Eq. 2 as described in [62,65]. Briefly, d(ln(F(R)h)/d(h) vs. h thus shows a maximum, which can be used as an approximated value for E g . The exponent n can be obtained as an inverse slop in the plot of ln(F(R)h) vs. ln(h  E g ) (Fig. S3, supporting information). The exponent n was found to be  0.5 (indirect allowed transition) for both the calcined and the sintered samples.
For the nano-sized powder calcined at 350 °C, the band gap was calculated as 1.93(2) eV which slightly decreases with increasing crystallite size to 1.81(5) eV at a calcination temperature of 1000 °C. After sintering at 1050 °C the band gap is 1.77(3) eV and decreases to 1.60(3) eV at 1250 °C (Fig. 13). The reduction of the band gap with rising sintering temperature is most likely due to the loss of lithium. Whereas, the higher band gap energy for the nano-sized samples compared to the sintered ones (bulk) is due to the well-known size effect [66,67].  Evolution of the crystallite size with the calcining temperature. The right scale shows the development of the specific surface area.

Fig. 3
Bulk densities of ceramic bodies after sintering at various temperatures (soaking time 1 h, heating rate 5 K min 1 ). Error bars correspond to the size of the symbols.  Thermogravimetric measurement (rate 5 K min 1 ) of a Li 0.5 Fe 2.5 O 4 compact in flowing air.
The inset shows the total weight loss and the calculated lithium loss after 1 h sintering.      Dependence of the enthalpy change ( trs H) of the order  disorder phase transition on the thermal treatment. The inset shows the phase transition temperature versus thermal treatment.

Fig. 12
Dilatometric measurement in flowing air of a ceramic body sintered at 1100 °C during the order  disorder phase transition (cooling curve, rate 5 K min 1 ).

Fig. 13
Indirect-allowed band gap energies for samples calcined between 350 and 1000 °C and ceramics sintered between 1050 and 1250 °C. The inset shows (F(R)·h) 0.5 versus h for a sample calcined at 1000 °C for 2 h. The uncertainties ( 0.05 eV) of the band gap values are smaller than the symbol sizes.