Background

Due to remarkable properties such as excellent cycle life, high thermal resistance, no memory effect, and low self-discharge, Li-ion batteries have attracted great attention in electrochemical energy storage [1–3]. More recently, the miniaturization of these power sources has been investigated to meet many applications in microelectronics such as real-time clocking (RTC), radio frequency identification (RFID), sensors, medical implants, memory backup power, solar cell, and smartcards [4–6]. One of the major challenges in the microbattery field is related to the manufacturing process and its compatibility with the integrated circuit technology. Particularly, metallic Li which is currently used as a negative electrode is not compatible with the reflow soldering process because of its low melting point (180.5 °C). Therefore, the rocking chair technology involving stable materials must be explored. Many researches focused on the development of new components with nanostructured materials and 3D designs [7–11].

As potential cathode materials, the NASICON-type compounds A3Fe2(XO4)3 (A = Li, Na; X = P, As, S) with 3D frameworks like Li3Fe2(PO4)3 have attracted considerable attention [12]. NASICON-type Li3Fe2(PO4)3 can generate 2.8 V vs. Li with an excellent capacity retention, and up to 2 mol of Li+ can be reversibly intercalated into Li3Fe2(PO4)3, delivering a capacity of 128 mAhg−1. NASICON has also a relatively high ionic conductivity resulting from the disorder of lithium ions in the structure, favorable redox properties, low cost, structural stability, and simple fabrication procedures [5, 6, 12]. Many methods have been reported to synthesize Li3Fe2(PO4)3 nanostructures such as hydrothermal [13, 14], gel combustion [15], solid state [16–19], solution with the use of citric acid [20], sol-gel [21], ultrasonic spray combustion [22], and ion exchange with molten salt [23, 24].

In this work, we report the fabrication of porous NASICON-type Li3Fe2(PO4)3 thin film electrodes by radio frequency sputtering. Sputtering techniques are widely employed in microelectronics for the successive deposition of thin films; they are the main process used to fabricate planar all-solid-state Li-ion microbatteries [25, 26]. To the best of our knowledge, this approach has never been applied to deposit Li3Fe2(PO4)3. We show that this material tested as a cathode reveals a good electrochemical behavior for all-solid-state Li-ion microbatteries.

Methods

Preparation of Li3Fe2(PO4)3 Thin Films

Firstly, thin films of LiFePO4 were deposited on titanium foil as substrate by radio frequency (RF) sputtering (PLASSYS). The target was LiFePO4 (purity 99.9 %, Neyco). The process was carried out in a vacuum chamber with a base pressure before deposition of 10−6 Torr. The sputtering gas was pure argon, and the working pressure was 10 mTorr with a gas flow of 21.5 sccm. A sputtering power of 150 W was applied to the target. The deposition time was 3 h. Secondly, the as-deposited thin films were annealed in air at 700 °C for 3 h with a heating rate of 10 °C min−1 (furnace: NABERTHERM Controller B 180).

Structural Characterization

Phases and crystallinity of thin films were examined by X-ray diffraction (XRD) with Cu Kα radiation (wavelength = 1.5405 Å) using a D5000 BRUKER-SIEMENS diffractometer. The morphology of the Li3Fe2(PO4)3 was investigated by scanning electron microscopy (SEM, Hitachi, S-570). Thickness measurement was performed by profilometry analysis using a DEKTAT XT (BRUKER) equipment. A portion of the sample is masked during deposition, and the thickness is given by the height of the step.

Electrochemical Measurements

The electrochemical tests of the half-cell were performed using Swagelok cell assembled in a glove box filled with purified argon in which moisture and oxygen content were less than 0.5 ppm. The half-cells consisted of metallic Li as counter electrode. Two circular sheets of Whatman glass microfiber soaked with lithium hexafluorophosphate in ethylene carbonate and diethylene carbonate (1 M LiPF6 in (EC:DEC) 1:1 w/w) were used as separator. Cyclic voltammetry measurements were performed using a Versastat potentiostat (AMETEK) in the range of 1.25–5 V vs. Li/Li+ with a scan rate of 1 mVs−1. Galvanostatic cycles were performed with a VMP3 (Bio Logic) at 25 °C between 2 and 4 V vs. Li/Li+. The different cycles of the charge/discharge rate were investigated at a kinetic rate of C/10 and C/5 considering that the theoretical capacity of Li3Fe2(PO4)3 is 128.25 mAhg−1. Hence, the applied currents were 3.3 and 6.6 μA cm−2, respectively. No additives or binder was used for all electrochemical tests.

Results and Discussion

Prior to battery assembly, the crystallinity of the thin films were examined by XRD. The XRD patterns of initial target and as-deposited sample are shown in Fig. 1. It can be seen that the target is well-crystallized and corresponds to the Olivine-type LiFePO4 phase (JCPDS file no. 040-1499). Various deposition parameters (such as target power, substrate temperature, or argon pressure) were previously investigated, and in all cases, the as-deposited thin films were amorphous. As it can be seen in Fig. 1, the annealing treatment leads to the crystallization of NASICON-type Li3Fe2(PO4)2 phases (JCPDS file no. 047-0107), which is known to have better electrochemical properties [27]. Three small peaks are attributable to the Fe2O3 phase. The oxidation state of Fe in LiFePO4 and Li3Fe2(PO4) are +II and +III, respectively. It can be noted that the annealing in air of amorphous LiFePO4 not only leads to crystallization but also to the iron oxidation according to (Eq. 1) [28, 29]:

Fig. 1
figure 1

XRD patterns of samples a LiFePO4 target, b as-deposited Li3Fe2(PO4)3, and c annealed Li3Fe2(PO4)3 at 700 °C for 3 h

Full size image
$$ 12{\mathrm{Li}\mathrm{FePO}}_4+3{\mathrm{O}}_2=4{\mathrm{Li}}_3{\mathrm{Fe}}_2{\left({\mathrm{PO}}_4\right)}_3 + 2{\mathrm{Fe}}_2{\mathrm{O}}_3 $$
(1)

However, the conversion reaction of Fe2O3 with Li+ occurs at low potential which is not interesting for its use as a cathode [30, 31].

The morphology of the Li3Fe2(PO4)3 thin films before and after annealing was observed by SEM. From Fig. 2a, it is apparent that the as-deposited Li3Fe2(PO4)3 layer is homogeneous with a thickness of 850 nm according to the surface profilometry analysis given in Fig. 2b. After annealing, the film became rough and a sponge-like texture clearly appeared (Fig. 2c, d). Hence, the thermal treatment promoted the formation of a mesoporous material with a larger surface area than the as-deposited layer.

Fig. 2
figure 2

a as-deposited Li3Fe2(PO4)3 with b profilometry analysis, low (c) and high magnification (d) SEM images of porous Li3Fe2(PO4)3 deposited by RF sputtering after annealing treatment

Full size image

The electrochemical characterization of the porous Li3Fe2(PO4)3 thin film was carried out by cyclic voltammetry (CV). Figure 3 shows the cyclic voltammogram (CV) is obtained. In agreement with literature, the electrochemical reactivity of crystalline Li3Fe2(PO4)3 is confirmed by the presence of anodic and cathodic peaks [3, 32]. These two peak pairs can be attributed to the reversible insertion reactions of two Li+ according to Eqs. (2) and (3):

  • First reduction (R1 2.68 V/Li) and oxidation peaks (O1 2.58 V/Li)

$$ {\mathrm{Li}}_4{\mathrm{Fe}}_2{\left({\mathrm{PO}}_4\right)}_3={\mathrm{Li}}^{+}+{\mathrm{Li}}_3{\mathrm{Fe}}_2{\left({\mathrm{PO}}_4\right)}_3+{\mathrm{e}}^{-} $$
(2)

  • Second reduction (R2 2.72 V/Li) and oxidation peaks (O2 2.85 V/Li)

$$ {\mathrm{Li}}_5{\mathrm{Fe}}_2{\left({\mathrm{PO}}_4\right)}_3={\mathrm{Li}}^{+}+{\mathrm{Li}}_4{\mathrm{Fe}}_2{\left({\mathrm{PO}}_4\right)}_3+{\mathrm{e}}^{-} $$
(3)
Fig. 3
figure 3

CV curve of the annealed Li3Fe2(PO4)3 thin film deposited on Ti substrate recorded at a scan rate of 1 mVs−1

Full size image

The presence of an additional reversible peak (O3 and R3 peaks) at around 2.4 V reveals that Li+ can also react with another phase. This active material is supposed to be LiFePO4(OH) into which lithium intercalation occurs through the reduction of Fe3+ to Fe2+ [33–35]. Compared with literature, it can be noticed that the peaks are quite broad. This phenomenon can be explained by the presence of pores. Indeed, the large surface area offered by the porous texture promotes the storage of charge at the surface according to a non-faradaic process. Hence, the contribution of the capacitive effect leads to the broadening of the peaks.

The galvanostatic charge/discharge profiles obtained at C/10 of the half-cell battery are shown in Fig. 4a. In agreement with the CV curve, the presence of three pseudo-plateaus confirms that Li+ ions react with Li3Fe2(PO4)3 and LiFePO4(OH).

Fig. 4
figure 4

Battery cycling tests at the kinetic rate of C/10 in the potential window of 2–4 V (a). Discharge capacity of cathode at multi-C-rate (b)

Full size image

The capacity values obtained for the as-deposited sample (Fig. 4b) are very low in the range of 1–2 μAhcm−2. But, a first discharge capacity of 88 mAhg−1 (23 μAhcm−2) is achieved at C/10 for the annealed sample. An average value of about 65 mAhg−1 (17 μAhcm−2) is delivered after 20 cycles. This capacity corresponds to more than 50 % of the theoretical capacity of Li3Fe2(PO4)3 (33 μAhcm−2). We can observe that capacity is slightly fading during the first cycles. The capacity lost may be caused by the different effects which are associated with side reactions due to the phase changes in the insertion electrode, dissolution of the active material, or the decomposition of the electrolyte. The capacity loss has been also discussed by Song et al. [36] who reported that capacity fading can be attributed to surface impurities (such as lithiated iron oxide which is also involved in the lithium insertion/extraction) leading to the irreversible reaction of Li+ with iron oxide and the formation of a solid electrolyte interface (SEI) at the surface of the electrode.

In order to study the operational stability of the NASICON-type Li3Fe2(PO4)3, the cycling life performance has been also studied at two different C-rates (Fig. 4b). From this graph, it can be seen that the capacity becomes stable after 10 cycles. At C/5, the battery capacity is about 50 mAhg−1 (13 μAhcm−2) and remains constant over 40 cycles. These results also show that the porous geometry of the mesoporous electrode is beneficial for tolerating the volume expansion that generally occur during alloying/de-alloying reactions.

Conclusions

In this work, we report the fabrication of mesoporous NASICON-type Li3Fe2(PO4)3 thin film by radio frequency sputtering. The electrochemical studies suggest that after annealing, the crystallized layer is also composed of LiFePO4(OH), which is able to react reversibly with Li+. In the first cycle, the discharge capacity reaches about 88 mAhg−1 (23 μAhcm−2) at C/10 and attains 65 mAhg−1 (17 μAhcm−2) after 10 cycles. The electrochemical characterizations also reveal a good stability suggesting that this material can be used as a cathode for Li-ion microbatteries.