Negative electrode materials for lithium-ion solid-state microbatteries

L. Baggetto

Research output: ThesisPhd Thesis 1 (Research TU/e / Graduation TU/e)

1434 Downloads (Pure)


Electronic portable devices are becoming more and more important in our daily life. Many portable types of electronic equipment rely on rechargeable lithium-ion batteries as they can reversibly deliver the highest gravimetric and volumetric energy densities. Lithium-ion batteries are currently rapidly expanding into very large-scale applications, such as hybrid (electrical) cars. Miniaturized autonomous devices, at the other end of the ‘spectrum’, are also becoming increasingly important. Characteristic for small autonomous devices is that they have to operate independently. When devices are becoming smaller and smaller it becomes, however, much more complicated to assemble batteries from their individual components. In addition, the contribution of inactive overhead mass and volume by, for example, the package will increase significantly. As the energy consumption for autonomous devices will be relatively small this opens up the possibility to integrate microbatteries directly onto electronic chips. Moreover, as certain applications have stringent safety requirements, for instance medical implants, integrated batteries ideally should not contain any hazardous liquids that can induce dangerous leakage issues. All-solid-state planar microbatteries are good candidates to serve as energy sources for autonomous systems as they can be integrated into microchips and they do not present safety issues arising from the use of a liquid electrolyte. Such batteries already exist in the pilot production phase and are usually produced using Physical Vapor Deposition techniques (PVD), such as magnetron sputtering and evaporation. However, a relatively low volumetric storage capacity of about 50 ¿Ah per micron cathode material thickness and per cm2 footprint area is generally achieved. Yet, it is still not sufficient to power future autonomous devices. In addition, metallic Li is unfavorable due to its very low melting point of 181 °C. Indeed, the reflow solder process widely used in the microelectronic industry is generally applied at higher temperatures. To prevent the use of pure Li metal, it would be better to make use of other negative electrode materials. In addition, ways to increase the energy density of solid-state microbatteries should be investigated. This could be achieved by, for instance, using the third dimension of the battery substrate material. The concept of 3D-integrated all-solid-state microbatteries has been presented (c.f. Chapter I). It consists of the step conformal deposition of Li barrier, negative electrode, solid-state electrolyte, cathode and current collector layers inside high aspect ratio features such as trenches and pores. The deposition of the battery materials should be achieved by means of techniques capable of covering highly structured substrates with layers of homogeneous thickness and composition. Low Pressure Chemical Vapor Deposition (LPCVD) and Atomic Layer Deposition (ALD) are suitable techniques to that purpose. Compared to conventional planar solid-state batteries this concept offers several advantages.The 3D-structuring of the substrate by means of anisotropic etching provides large surface area enlargement factors, which is expected to increase the effective storage capacity by more than one order of magnitude. Moreover, the use of Li-alloying materials to replace pure metallic Li as negative electrode ensures compatible IC integration. As a first step towards 3D-integrated batteries, the investigation of potential Li barrier layers has been conducted (c.f. Chapter III). By analogy with Cu barrier layers, layers of 70 nm TiN, Ta and TaN, deposited by means of magnetron sputtering, have been investigated. The electrochemical results suggest that TiN is most suitable as a Li barrier layer since it only absorbs less than 0.02 Li per TiN formula unit. Thus, TiN has been applied extensively as a barrier layer and current collector for the study of potential negative electrode candidates. Moreover, a detailed comparison of the electrochemical response and crystallographic preferred orientation of sputtered and ALD TiN films has been presented. ALD TiN shows better barrier properties, which is perhaps related to its strong preferred orientation of the {200} planes parallel to the sample surface or to the larger size of its crystallites. The storage mechanism of Li is not elucidated but perhaps occurs as a solid solution of Li within TiN or at the grain boundaries of TiN crystals. Ex situ X-Ray Diffraction (XRD) measured after electrochemical cycling has not shown significant changes in the TiN crystallographic structure, which further confirms that the material does not react significantly. ALD is a suitable technique for growing step-conformal thin layers into 3D-structured substrates, which makes the technique suitable for creating 3D negative electrode stacks (c.f. Chapter V). The reaction mechanism of Si electrodes has been described based on several literature reports (c.f. Chapter IV). In situ XRD clearly evidences the transformation of crystalline Si into an amorphous material during the initial Li-ion insertion, which is ultimately completed by the crystallization of the electrode material into metastable cubic Li15Si4. During Li-ion extraction, an amorphous material is created from the latter crystallites and the formation of amorphous Si as the fully delithiated electrode material is evidenced. During the subsequent cycles, the amorphous material reversibly transforms via two slopes, or quasi-plateaus, until the crystallization into Li15Si4 occurs. As a consequence, Si shows very high practical gravimetric and volumetric capacities, i.e. 3579 mAh·g-1 and 8303 mAh·cm-3, respectively, assuming complete conversion into Li15Si4. In situ Nuclear Magnetic Resonance (NMR) results show that the formed Si-Si and isolated Si atoms configurations are different from those of the Li-Si phase diagram compounds. Both configurations are created during the initial stage of Li-ion insertion and increase upon further insertion until only isolated Si atoms are formed. LPCVD poly-crystalline Si (poly-Si) thin films of 50 nm thickness exhibit very favorable thermodynamic and kinetic properties as evidenced by quasi-equilibrium, impedance spectroscopy and rate capability measurements. The cycle life of the electrode has been investigated in two liquid electrolytes and when the electrode is covered by a solid-state electrolyte film. The continuous decomposition of the electrolyte resulting in the formation of a thickening SEI layer is expected to be responsible for the poor capacity retention of the electrodes. In contrast, poly-Si thin films covered by a solid-state electrolyte film can be cycled 2000 times without capacity loss. Evaporated Si thin films have been used in order to investigate the impact of solid-state electrolyte sputter-deposition. The formation of an interlayer material has been evidenced by transmission electron microscopy (TEM). The thickness and composition of this material depends on the sputter deposition power and gas composition, and results in a different electrochemical activation of the electrode/electrolyte stack. The charge transfer kinetics of activated electrodes has been investigated as a function of temperature. As expected, Arrhenius laws have been found for the various evaporated Si electrode/electrolyte systems and the corresponding activation energies have been determined. The huge volume changes of Si upon (de)lithiation can limit the amount of active electrode material and hence its storage capacity. Using larger layer thickness will obviously increase the total storage capacity of the electrode but it will also result in poorer capacity retention due to mechanical delamination or pulverization of the film. Limiting the extent of Li-ion insertion or extraction can prevent this detrimental effect, however, this is achieved at the cost of storage capacity. In order to increase the storage capacity while maintaining small expansion/shrinkage distances, 3D-nanostructured Si electrode systems have been investigated (c.f. Chapter V). In this way, not only the absolute changes in volume are limited but the diffusion length is also kept low to ensure fast diffusion kinetics. Three types of 3D-nanostructured Si electrode systems have been investigated, i.e. thin films deposited inside high aspect ratio pores and trenches, nanowires and honeycombs. The electrochemical performance and morphological changes of these structures are described in detail. The deposition feasibility of a highly structured negative electrode stack to be applied in future 3D-integrated batteries has been demonstrated. The stack comprises a TiN thin film deposited by ALD serving both as current collector and Li barrier layer, which is covered by a poly-Si thin film as negative electrode material. In comparison with planar films, these poly-Si films present a storage capacity increase of about 5 for the highest aspect ratio electrodes. Poly-Si deposited inside pores show an ‘island-like’ structure, which most likely results from the surface morphology of the underlying material. It has been found that the step coverage of poly-Si can be considerably improved by growing TiN and poly-Si inside wide trenches, which results in the growth of much smoother poly-Si films. Further optimization of the deposition conditions and 3D geometry can easily result in a storage capacity increase of more than one order of magnitude with respect to planar films. As found for planar thin films, the electrode cycle life appears to be limited by the continuous growth of the SEI layer inside the pores. Hence, the use of a solid-state electrolyte should be highly beneficial for enhancing the electrode cycle life, as has already been demonstrated for planar poly-Si thin film electrodes. Si nanowires present a poor initial electrochemical activity when evaluated with Cyclic Voltammetry (CV). Employing constant (dis)charging low currents ensures that the material gets activated. The reasons for the initially sluggish electrode kinetics are most likely resulting from the high electrical resistance of the nanowires, in combination with a poor wetting of their surface. A poor surface wetting reduces the surface area available for the charge transfer reaction and also increases the diffusion distance of Li ions. The morphological changes of the nanowires after cycling show that they tend to agglomerate upon expansion/shrinkage to form a porous network separated by voids, which might not be favorable if the electrode is to be repeatedly cycled for a long time. The electrochemical performance and morphological changes of honeycombnanostructured Si have been investigated. Si honeycomb arrays present a moderate cycle life, which can be improved by limiting the (de)lithiation reaction. The morphological changes of the structure have been investigated and reveal that the honeycomb structure can undergo very large deformations. The analysis of the dimensional changes as a function of Li content reveals that the structure does not deform isotropically. However, the total volume varies linearly and reversibly as a function of Li content. The degradation mechanism has been identified and results from the detachment of small parts of the honeycombs from the current collector after repeatedly cycling. The detachment is mainly caused by the wall bending and subsequent cracking, and the wall thickening at the isolated triple points which results in their mechanical detachment. Ge electrodes have received little attention in the existing literature because of the high cost of the material. For the application in thin-film batteries, however, the cost related to the amount of employed material is not a critical issue. Moreover, Ge presents several advantages over Si, such as a two orders of magnitude higher diffusivity for Li ions and four orders of magnitude higher electrical conductivity. The electrode reaction mechanism is not known and has been investigated in detail using electrochemical methods, in situ XRD, in situ X-ray Absorption Spectroscopy (XAS) and ex situ X-ray Photoelectron Spectroscopy (XPS) (c.f. Chapter VI). The thermodynamic and kinetic properties of Ge thin films clearly evidence that this material is a suitable negative electrode candidate for Li-ion microbatteries. The material shows a large storage capacity, a high rate capability, resulting from both fast charge transfer and solid-state diffusion kinetics, and favorable thermodynamics. The in situ XRD investigation has shown that lithiated Ge thin film electrodes remain amorphous until crystallization in cubic Li15Ge4 occurs. This corresponds to a storage capacity of 1385 mAh·g-1 or 7366 mAh·cm-3 of starting material. Upon Li-ion extraction, a reversible process is found. In order to get more insight in the structure of the amorphous material formed prior to cubic Li15Ge4, in situ XAS has been conducted. The corresponding results indicate that the electrochemical reaction proceeds with the reversible formation of short range ordered LiGe, followed by the formation of a material with increasing Ge-Ge interatomic distance and decreasing coordination number. Near the end of the reaction, the local ordering approaches that of Li7Ge2 and finally complies with the short range ordered structure of Li15Ge4 at full lithiation. Finally, XPS surface measurements indicate the formation of Li2CO3 and (PPO)n as the main constituents of the top surface of the SEI layer formed onto Ge electrodes cycled in 1M LiClO4 in PC electrolyte. As already discussed, alloying negative electrode materials (e.g. Si, Ge and Sn) undergo tremendous volume expansion, which can be detrimental for the material lifetime that is often reduced to a few cycles for thick electrodes. Instead of using the pure element, oxides and nitrides can be employed as they are expected to present improved cycling performance. The generally-accepted scenario to explain the favorable cycle life of these electrodes is a conversion reaction during which an inert nanostructured lithia (Li2O) or lithium nitride (Li3N) matrix is irreversibly formed concomitantly with the pure element during the initial Li-ion insertion. These matrices are expected to accommodate part of the stress associated with the large volume expansion/contraction resulting from the insertion/extraction of Li ions into/from the active elemental nanoclusters. In the case of tin nitride, the conversion mechanism and the electrochemical properties of the material have not been thoroughly investigated thus far. Consequently, a careful investigation has been conducted for this material (c.f. Chapter VII). The electrochemical behaviors of pure tin and of tin oxide thin films have been given as references. The characterization of as-prepared tin nitride thin films with layer thickness ranging from 50 to 500 nm and compositions of 1:1 and 3:4, as well as their electrochemical responses have been presented. Using Rutherford Backscattering Spectrometry, the amount and composition of the starting tin nitride material was determined, which allowed the calculation of the quantity of Li reacting (ir)reversibly with the electrode material. These results have shown a large discrepancy with respect to the expected reaction mechanism. Thus in situ XRD has been performed in order to characterize the structure of the converted material for different Li contents, however, not successfully. Mössbauer spectroscopy (MS) has been employed on thicker layers aiming at probing the chemical environment of Sn atoms as a function of Li content. XRD results on the as-prepared thick films of composition 1:1 indicate the presence of an amorphous or nanocrystalline structure in which nanocrystals of Sn3N4 and of, most likely, reacted or distorted ¿-Sn are embedded. According to MS results, the amorphous material appears to be a mixture of tetrahedrally- and octahedrally-coordinated Sn4+, in combination with Sn2+ species which most likely result from N deficient octahedral and tetrahedral sites, especially octahedral. The films of composition 3:4 are composed of an amorphous material which continued a crystalline growth after 300-500 nm from the interface with the TiN-covered substrate. According to XRD results, the material mainly consists of the spinel Sn3N4, and minor traces of reacted ¿-Sn and orthorhombic SnO2. The MS signature of this material shows the response of Sn4+ located in octahedral and tetrahedral sites of the spinel structure while no ¿-Sn or SnO2 was detected. At 77 K, however, the presence of Sn2+ was revealed as a minor secondary constituent. These species are most likely located in the amorphous material observed with TEM and are similar to the Sn2+ species measured for the 1:1 material. When these tin nitride materials are inserted with Li-ions, large modifications of the structure occur. The 1:1 material transforms into Li3N and intermediate Li-Sn phases and ultimately reaches the composition ‘Li22Sn5’. During Li-ion extraction, modified Sn4+ environments are formed while the Li content in the LiySn clusters decreases. At full delithiation, the electrode shows a signature for modified Sn4+ octahedral ad tetrahedral environments together with a mixture of LiSn and Li7Sn3. The presence of Li7Sn3 as a minor product is most likely related to the impossibility to fully remove Li ions from the thicker electrode film, as was also found for the thick films of composition 3:4. Thus, it can be assumed that the final composition in the Sn-rich cluster should be close to LiSn. For the 3:4 material, the initial insertion of Li ions is different from that of the subsequent cycles, which is most likely due to the co-existence of original Sn4+ octahedra and tetrahedra with Li3N and LiySn products. During the first extraction of Li ions, however, the electrode potential profile is very similar to that measured during the subsequent cycles or to that of the 1:1 material, which is most likely indicative of similar reactions. During the initial lithiation, the preferential consumption of octahedral Sn4+ sites is measured. As N is linked to 3 Sn in octahedral and 1 Sn in tetrahedral sites, it can be assumed that the reaction of Li with N will preferentially affect the octahedra. The composition of the LiySn products is not elucidated and it seems that specific sites of Li-Sn phases are formed. At full lithiation, similarly to what is obtained for the 1:1 composition the signature for ‘Li22Sn5’ is measured. During Li-ion extraction, the isomer shift associated with LiySn products increases and modified Sn4+ octahedral and tetrahedral environments are formed. At full delithiation, a Li-Sn product with the isomer shift of LiSn is measured. Its quadrupole splitting, however, is substantially higher than the value expected for LiSn. The much higher quadrupole splitting most likely indicates the electrode nanostructuring and also the formation of either a specific site of LiSn or of a complex ternary compound
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Chemical Engineering and Chemistry
  • Notten, Peter, Promotor
  • Roozeboom, Fred, Promotor
  • Hintzen, Bert, Copromotor
Award date5 Jul 2010
Place of PublicationEindhoven
Print ISBNs978-90-386-2284-2
Publication statusPublished - 2010


Dive into the research topics of 'Negative electrode materials for lithium-ion solid-state microbatteries'. Together they form a unique fingerprint.

Cite this