Community Research and Development Information Service - CORDIS

FP7

SUPERLION Report Summary

Project ID: 214832
Funded under: FP7-NMP
Country: Sweden

Final Report Summary - SUPERLION (Superior Energy and Power Density Li-Ion Microbatteries)

Executive Summary:
SUPERLION passed through an “academic” phase (months 1-18), dominated by basic research efforts by the University partners, prior to the participation of industrial partners (months 19-36), aimed at providing “proof-of-concept” through the fabrication of one or more functioning three-dimensional 3D-MBs.
• 3D-MB design specifications: Area Gain (A.G.) factor 30; benchmark “footprint” capacity: 30-50 ?Ah/mm2 - delivering an energy density of 100-200 ?Wh/mm2 for a stack-thickness of 500-1000 ?m.
• FEA modelling exposed significant non-uniformities in current distributions within working cells, which were strongly dependent both on design architecture and choice of component materials.
• Deposition of negative and positive electrode materials as conformal films on nano-structured current collectors gave many new advances: LiFePO4 onto the surface of reticulated vitreous carbon; a new “copper-trench” current-collector architecture; thin-film polymer electrolytes self-assembled as surfactant oligomers on LiFePO4 particles.
• Negative Sb and Cu2Sb electrodes have been electrodeposited successfully onto 15µm Cu micro-trenches. Likewise, LiFePO4 and LiCoO2 positive electrodes have been deposited onto columnar and needle-like Al current-collectors to give good capacity retention at 1C rate. CuS cathodes (with a perforated Si substrate) cycled more than 400 times with ~0.1%/cycle capacity-loss in the range 0.1-2.0 mA/cm2 with pulse-power capability ~ 160 mW/cm2. Corresponding LiFePO4-cathode cells gave a capacity of 80 µAh/cm2 with an average potential of 3.4 V after 20 cycles, and a pulse-power peak of ~ 175 mW/cm2 for a 1s discharge pulse-step. MnO2 half-cells showed capacities (in mAh/cm2) up to 200 times greater than for comparable 2D planar electrodes. Charge/discharge behaviour for LiFePO4 shows a significant increase in capacity in going from using 5 (130 ?Ah/cm2) to 10 (870 ?Ah/cm2) wt.% inks.
• “Trench-type” 3D-MB cells have been developed by dropwise insertion of LiCO2 into the trenches, and by attachment of the composite cathode to the anode/polymer structure through the application of gentle pressure. LiFePO4-based 3D-MB half-cells have been assembled successfully with aperiodic architecture to give a capacity of 1.5mAh/cm2. Complete <TiO2/solid polymer electrolyte/LiFePO4> 3D-MB whole-cells with aperiodic architecture gave a stable OCV, while concentric Li/LiFePO4 half-cells gave a capacity of 1.5-2.5 mAh/cm2 for >100 cycles and >200mW/cm2 pulse-power capability.

Proof-of-concept activities by the industrial partners focussed on three test-cell types: a LiFePO4-based flexible flat cell; MC 614 and V500 coin-cells. To summarize the findings of this work: although these showed promising characteristics in terms of size and energy-density, their electrochemical performance must be further improved for use as realistic power sources in future MEMS or medical devices. Improvements must be made primarily in the materials processing phase; battery production processes must also be optimized, since the intrinsic performance of single-cell components has been well demonstrated on a laboratory scale. Rate capability and pulse performance must also be improved. Internal resistance should also be further reduced. Higher system capacities will also be necessary to increase their capacity range from µAh to mAh. A higher volumetric energy density will also be required in some applications, especially when higher capacities are needed as the cells get larger.
The feasibility of incorporating 3D-MB concepts into conventional Li-ion battery production was seen an extra bonus of the project. The most interesting candidate was adjudged on the basis of its superior mechanical properties to be the “aperiodic sponge” architecture. However, electrochemical stability limitation of the Ni-foams is still seen as a problem. Financial aspects must also be carefully evaluated.
By the end of the project, proof-of-concept for 3D-MBs has been achieved at a design level, but no pilot-line production of 3D-structured electrode materials or cells yet exists.

Project Context and Objectives:
Access to on-board micro-battery power is fast becoming an essential requirement in many of today’s emerging technologies. It is clear to all that down-scaling in the micro-electronics industry has far outpaced advances made in small-scale electrical power supplies. Indeed, the continued absence of efficient on-board power is already becoming a severe hinder to further advances in many critical areas; typically, in many MEMS-type devices, not least biomedical in vivo micro-machines. Down-scaling the battery power source clearly presents a serious challenge to Science, but the advent of nano-materials and nano-structures now places new resources at our disposal to facilitate a fresh attack on the problem. SUPERLION make this attack.
More specifically, recent miniaturisation of electrical, mechanical and optical systems has led to a variety of devices with a total volume of little more than 10 mm3. Though many have never even heard of these, there is an emerging range of micro-scale devices (called Micro-Electro-Mechanical Systems – MEMS) which are destined to change our lives completely. These devices include smart micro-sensor arrays for detection of hazardous materials, anti-terrorism microchip sensors, self-powered unmanned air micro-vehicles, fully integrated RF identification cards, non-volatile PC memory backup, not to mention the exciting new area of biomedical in vivo micro-machines, such as pacemakers, defibrillators, neural stimulators, drug delivery systems, micro-imaging capsules, etc. The USA demand for cardiac and orthopaedic implants, for example, is increasing at ca. 11% annually.
As intimated above, the major obstacle today to the widespread development of these micro-devices is the lack of a suitable autonomous battery power source. The realization that the power requirements of such micro-systems cannot be met by existing lithium or lithium-ion solid-state thin-film batteries in flat 2D-configuration has led to the search for a true 3D-microbattery(3D-MB), where the critical bottleneck is the shortage of cheap, light and powerful micro-/nano-fabrication materials. The fundamental problem with planar 2D thin-film batteries is their limited active electrode surface. Moving into the 3rd dimension invokes the simple principle of geometrical Area Gain (A.G.), resulting in electrode volume gain with associated increase in battery capacity. An obvious approach to achieve significant A.G. (by a factor of up to 25-40) is to replace a continuous by a perforated substrate (with high aspect-ratio channels), thereby better utilizing the dead volume of the substrate.
While focussing on this central goal of achieving greatly improved MB performance through the use of new nano-materials and -architectures in a 3D-MB concept, it is tempting to raise the question as to what extent these same (or related) techniques can also be used to improve the performance of the conventional normal-scale Li-ion battery. Can we, indeed, move from today’s methodologies involving an assembly of five separate cell components to a fabrication technique involving one single chemically-formed battery laminate, comprising interpenetrating (“nano-architectured”) substrates, anode and cathodes elements separated by an electrolyte? Can our 3D-MB related work also serve as a stimulus to a paradigm shift in the production of conventional rechargeable Li-ion batteries, which could meet current demands for better large-scale battery performance in, typically, EV/HEV applications?

The MAIN PROJECT OBJECTIVES of SUPERLION are therefore:
• Synthesis and fabrication of novel nano-architectured battery materials and components for use in 3D-MB concepts.
• Implementation of these in fully integrated thin-film 3D-microbatteries (3D-MBs) with energy, capacity (and power) per footprint area at least an order of magnitude greater than that of planar (2D) thin-film battery.
• Implementation of at least some of these 3D-MB concepts in conventional normal-scale Li-ion battery fabrication.
• “Proof-of-concept” by showing that some 3D-MB concept from the project can power an electronic and/or a medical MEMS device.

The project, in this way, establishes a platform of 3D nano-architectures and micro-/nano-fabrication techniques, and the associated enabling Science for a new generation of microbatteries.

Project Results:
1. Modelling the 3D Microbattery
1.1 Static modelling of various 3D microbattery architectures
The main focus of the static simulation of the 3D Li-ion battery was to derive the capacities and energy limits for prospective 3D Micro-Batteries (MBs) of various designs. If 3D-MB design is based on conformal layer deposition, then the capacity and energy gain is a function of the geometrical Area Gain (AG). However, several prospective MB designs are essentially three-dimensional; their properties cannot therefore be calculated on the basis of AG alone. A special analysis must be made for such MB designs; namely, the capacity and energy of the 3D-MB is calculated on the basis of geometrical parameters and the density/specific capacity of the materials. This approach and details of the derivations involved can be found in the “Static Modelling” report placed on the SUPERLION website. The most relevant MB designs considered were: 3D interdigitated (row and interlaced), 3D continuous anode, and 3D continuous cathode.
The following questions relevant to 3D-MB design can be solved with the help of the delivered Matlab procedures: (1) What is the minimal thickness of the 3D-MB necessary to reach the targeted values of 300 µAh/mm2 in capacity and 900 µWh/mm2 in energy density? (2) What cathode and anode column radii are needed to reach optimal volume utilization? Precise answers to these questions depend on the design parameters and the electrode materials selected. Interlaced and row designs are shown to be identical in capacity and energy per footprint area; and device thickness of 1.3-1.5 mm is sufficient to reach the targeted capacity and energy densities per footprint area.
The second important 3D-MB design involves continuous (tube formed) anodes and cathodes. Continuous cathode and especially continuous anode geometries provide higher volume utilization than a 3D interdigitated design. Smaller device thickness (0.8-1.0 mm) can thus be achieved.
1.2 Dynamic modelling of 3D microbatteries
Dynamic modelling is essential in studying discharge processes in a 3D-MB, since it provides information on current densities within the battery, and hence on material utilization. Current moves differently through the battery depending on the 3D-MB architecture. In order to simulate the electrochemistry under operation of a complete 3D-battery cell, such a system needs to be modelled in all three dimensions. This has been accomplished using Finite Element Analysis (FEA) through the COMSOL Multiphysics program package. The input parameters to the program are here: the specific 3D-MB geometry, the electrochemical description, and material-specific parameters, such as conductivity.
The complete battery – involving electrochemical processes in both the electrodes and the electrolyte - has been modelled for both interdigitated “trench” and pillar designs. The FEA simulations of 3D-cells of graphite, 1 M LiPF6 in EC/DEC and LiCoO2 in the trench-model geometry were performed using concentrated solution theory, Maxwell-Stefan diffusion and Butler-Volmer kinetics. In addition to modelling the overall discharge characteristics, the model also exposed large non-uniformities arising in the current distribution. These were seen to be dependent both on architecture – trench depth, trench separation, plate shape – and on choice of material. Since non-uniform activity at the electrode/electrolyte interface results in a non-uniform utilization of the active material, these factors have a substantial influence on the charge/discharge profile of the batteries. A non-optimal battery design thus results in a non-optimal current distribution and electrode activity, and hence to an under-utilization of the active material. At the beginning of a discharge cycle, delithiation and lithiation of the electrodes starts directly from the plate tips in the trench-cell, leading to a fast depletion and accumulation of Li ions in these regions of the electrodes. This is due to the inhomogeneous current density distribution which is, in turn, caused by the electrode tip being considerably smaller than the corresponding surface of the opposite plate; this results in a concentration of current density at the tip – a direct effect of the interdigitated design. The higher local reaction rates at the plate tips will thus limit the current through the battery; the current has to find a different route through the cell, which brings the electrochemical processes to a premature end - at < 70% state-of-charge (SOC). In addition to this effect, it is anticipated that inhomogeneous expansion and contraction of the electrode material can result in cracking and disconnection to the electrodes, leading to poor capacity retention on cycling.
2. Electrode and electrolyte deposition processes
2.1 Current collectors
The key objective here is to deposit nanostructured current-collectors, and develop conformal and space-filling methods for subsequent deposition of the electrode materials; alternatively, to fabricate porous solid electrolyte containers as templates for subsequent deposition of both active electrode layers.
2.1.1 Fabrication of Cu and Ni nanorod arrays
Arrays of free standing copper nanorods were fabricated directly on commercial copper disk substrates by cathodic electrodeposition using a simple process developed earlier. Cu nanorods are grown from an electrolytic bath consisting of CuSO4 •5H2O 100 g L?1, (NH4)2SO4 20 g L?1 and diethyl-tri-amine 80mLL?1, inside the pores of an alumina membrane placed on top of the Cu substrate. Before electrodeposition, the Cu substrate was mechanically polished. After being rinsed with deionized water and ultrasonically cleaned in ethanol, the copper cathode was assembled in front of a copper anode, with the cellulose paper separator between the electrodes. The resulting two-electrode stack was kept under a constant pressure by using two stainless-steel clamps during the deposition process. Electrochemical Cu deposition was achieved using a pulsed cathodic current technique, with a two-step profile. The pulsed electrodeposition technique promotes grain nucleation and avoids diffusion limitations and is preferred over the constant current technique. After deposition, the two-electrode stack is removed from the solution under current, and the cell is dismantled. The cathode is soaked in hot alkaline solution (pH=14, 80 oC) to remove the AAO membrane, then in an acidic copper sulphate bath to dissolve the surface copper oxide. Current collectors are then stored into a glove box under a 6 atm. argon atmosphere preventing them from further oxidation. Uniform, defect-free and self-standing nanorod arrays are obtained which are directly attached to the underlying substrate. The length of the arrays could be varied with deposition time. With longer deposition times, the nanorod arrays tend to form bundles; judicious use of these could result in novel micro current-collectors.
2.1.2 Fabrication of Al nanorod arrays
Growth of Al nanorod current collectors has been achieved by direct electrodeposition of Al nanorods onto a planar aluminium substrate using porous alumina membranes from Ionic Liquids (ILs). Arrays of Al nano-rods (~200 nm diameter, ~5 ?m high) have been grown on Al substrates using ([EmIm]Cl)/aluminium chloride (AlCl3) (1:2 ratio) as the deposition electrolyte. The Lewis acid ionic liquid was prepared by slow addition of AlCl3 in [EMIm]Cl in an argon filled glove-box. The two compounds are both solid at room temperature and when mixed form a liquid, through a highly exothermic reaction. Al nanorods were directly synthesised, using the pulsed conditions. Free-standing nano-arrays of aluminium rods were obtained after dissolution of the alumina template in an aqueous solution of CrO3 (1.8 wt.%) and H3PO4 (6 wt.%).
Electrochemical cleaning of the surface of Al substrate with several potential scans by cyclic voltammetry prior to Al electrodeposition has been shown to be an effective step for achieving better adherence of the Al deposit on the substrate. This procedure reduces the thick Al2O3 layer present at the substrate surface and results in better adherence of Al nano-rod deposits. It has been possible to further improve the adherence of Al nanorod deposit, resulting in much better homogeneous arrays of nanorods, by slightly modifying the growth parameters, especially the pulsed potential steps. The goal was to increase the nucleation on the substrate and to avoid any diffusion limitation generally occurred during electrodeposition in such nanochannels.
2.1.3 Template–free electrodeposition of structured Al current collectors
Aluminium electro-deposits were grown on aluminium substrates from ionic-liquid electrolytic baths using pulse currents. The deposits had a mole-hill type morphology which was found to adhere nicely to the substrate. Differences were seen in deposition characteristics; typically in terms of particle-size and distance between particles.
Another common technique to deposit aluminium is the potentiostatic deposition method. Literature has often reported it as performed on several substrates and, each time, dense aluminium layers were obtained. A 2h deposition time was used in the [EMIm]TFSI/AlCl3 medium for various values of the imposed potential. It was found that low voltages allow dense layers as expected although increasing the voltage shows another aspect. Thicker aluminium layers were anticipated but, instead, the substrate was found not to be totally covered. This is explained by a partial decomposition of the electrochemical bath occurring at the same time. The partial decomposition seems to have a beneficial effect on the structuration of the aluminium deposits.
Literature reported that cetyltrimethylammonium bromide (CTAB) allowed a columnar growth of ZnO during hydrothermal and electrochemical syntheses. Coupling this additive with a high voltage could allow a good structure on our aluminium deposits. Experiment could confirm that the additive had no effect on the deposits obtained in [EMIm]TFSI/AlCl3/CTAB for -0.2, -0.5 and -0.7 V vs. AlIII/Al. A columnar aluminium deposit was first obtained without the help of any template at -1.0 V vs. AlIII/Al, with 0.5% mol CTAB in the electrochemical bath. Different textured aluminium current collectors were obtained using reduction processes in different deposition techniques. It was tempting to obtain a structure via the opposite route, i.e., by oxidation of an aluminium substrate. Different attempts were made; the most efficient was also the simplest: cyclovoltammetric oxidation. Different parameters had to be optimized like choice of the bath ([EMIm]TSFI/AlCl3), potential range (- 0.7 to 4 V vs. AlIII/Al), scan rate (100 mV s-1) and number of scans (100 and 1000). Each of these parameters was important to obtain a good oxidation of the substrate.
2.1.4 Silicon microchannels made by lithography
Philips Research produce perforated Si wafers: 6-inch in diameter, 700 ?m thick, N-doped and has a conductivity of about 0.1 S cm-1. The DRIE technique was employed to etch the holes in the wafer. The aspect ratio achieved was ca. 20:3. Cross-sectional SEM images show the holes to be 60 ?m in diameter and with 400 ?m penetration depth, with a ca. 10% broadening of the channel and narrowing of the inter-channel space, respectively. This is not critical, because the design of mask considers possible widening of the micro-channels during the etching process. The roughness of the internal surface is relatively low and is not expected to be detrimental for the deposition of successive battery layers.
2.1.5 Fabrication of high aspect-ratio perforated and interlaced Si substrates
A mask design comprising 42 chips of 11.74 mm x11.74 mm with a perforated pattern of 7x7 mm2 has been developed. The (round) chips have microchannels with a 50 ?m diameter and 30 ?m walls. The aspect ratio of the 350 ?m-thick substrate is 7 and the geometrical Area Gain (A.G.) factor is 9. The substrates were prepared by means of an inductively coupled plasma dry-etching technique (DRIE). Operation parameters such as flow rate, SF6 gas pressure, duration of etching-passivation steps and RF-power-coil power were adjusted to obtain optimal conditions for the microfabrication of the desired microstructures. A morphing procedure was applied based on controlling the duration of the etching step. The final etching process consisted of varying the length of the etching step in 500 deposition/etching cycles from 6 to 10 s. This suppressed the narrowing of the channels and reduced the roughness of the inner surface of the micro-containers formed. The mask design for the interlaced silicon substrates has also been developed. The pattern of interlaced micro-container arrays was prepared by double-side DRIE.
2.1.6 Conformal coating of the perforated Si substrate
After the DRIE perforation, the silicon wafers showed a significant difference between the surface morphology inside a microchannel and that of a flat area. Before carrying out electroless nickel or gold deposition, perforated samples were cleaned either in acetone and then in ethanol, or in cyclohexane followed by immersion in a solution of chromo-sulfuric acid. Modification of a metal-assisted etching method has been made to facilitate an appropriate roughness level of the Si to give strong adhesion to the current-collector. Two electrolytes, sulfate and sulfamate, were tested for the electroless deposition of thin-film nickel on the three-dimensional perforated silicon. The deposits were characterized in terms of their structure, morphology, crystallization behaviour, and composition. It was found that a sulfamate-based electrolyte with double complexing/buffering agents enables the deposition of a nickel layer with a homogeneous morphology and composition at a controllable reduction rate. The coating is composed of about 1.5 nm-size crystallites and contains 12-14% phosphorous.
2.2 Electrodes
2.2.1 Negative electrodes
2.2.1.1 Electrophoretic deposition of SnO2 nanoparticles on Cu nanorods
Electrophoretic deposition (EPD) technique has been used for the conformal deposition of SnO2 nanoparticles onto Cu nanorod current collectors. EPD is a versatile technique that allows deposition of nanomaterials with controlled size and crystallinity. Commercial SnO2 nanoparticles were used to prepare the dispersion for EPD in isopropanol. Polyethylene imide (PEI) was added as a steric stabilizer, to improve the stability of the suspension. As SnO2 particles showed a slight negative zeta potential in this media, a mixture of I2/acetone was added to the colloidal suspension to generate protons, thereby switching the zeta potential from negative to positive values. The composition of our ELD bath was SnO2 nanoparticles (0.5 g.L-1), PEI (0.05 g.L-1), I2 in acetone (a few droplets of a 1 g.L-1 solution) in isopropanol. The mixture was immediately used after sonicating for 25 min. Electrophoretic deposition of SnO2 nanoparticles onto Cu nanorods have been carried out in a homemade set up, with a stainless steel anode and Cu nanorod current collector as the cathode. Several parameters including solvents, concentration, applied field and time of deposition have been varied to obtain optimum deposition onto the Cu nanorod current collectors. A controlled field of 100 V/cm was applied between the electrodes. Some of the interspaces between the nanorods are observed to be filled with SnO2 nanoparticles. EPD deposition of SnO2 thin film has also been carried out on planar Cu substrates in order to compare the performance characteristics of 2D and 3D electrodes.
2.2.1.2 Electrolytic Sb for nanostructured Cu2Sb anodes
The electrodeposition technique has been chosen to prepare nanostructured Cu2Sb active material for 3D MBs, since Sb will be alloyed with Cu from Cu nanorod current collector. Conformal deposition of antimony (Sb) onto Cu nanorods is achieved, and the aim was thus to prepare Cu2Sb as active electrode material by alloying of the Sb with the copper current collector to prepare a 3D nanostructured negative electrode.
Antimony was deposited from an acidic solution of Sb-tartrate salts. The electrodeposition solution was based on a mixture of Sb and sodium tartrate, with the Sb concentration fixed to 0.15 M and the tartrate concentration set to 0.45 M. The pH of the solution was decreased to 1 by addition of sulphuric acid. The reduction of Sb(III) to Sb0 can then be written as:
[Sb2(C4H4O6)2]2+ + 6e- ? 2Sb0 + 2C4H4O62+
The Sb reduction potential in this solution was evaluated for reduction at a planar Cu subtrate and then reduction on a Sb layer previsouly deposited on a Cu substrate. A Pt disc was used as counter electrode and the reference electrode was a saturated calomel electrode. To promote the diffusion of the electroactive species within the 3D structure and thus to obtain a uniform coverage of the complex 3D surface of the Cu nanorod current collectors, the electrodeposition was performed using pulsed-current steps rather than a simple galvanostatic technique. 1st step: a constant current (idep) was applied for few ms (tdep) to favour nucleation. The electrodeposition parameters were varied to obtain optimum conditions. Homogeneous and conformal Sb deposits were obtained under the optimized conditions: a deposition current fixed to -1 mA.cm-2 was held for 10 or 50 ms and was followed by a rest step (0mA) of 50ms to allow diffusion of the Sb (III) complexes to the Cu 3D surface. When the deposition time and rest time were the same, larger Sb particles were obtained between the nanorods. The Sb deposit appears smoother when the deposition step was decreased from 50ms to 10ms. Both deposits can be suitable as 3D electrodes as they appear to cover the 3D nanorod current collector homogeneously.
2.2.2 Positive electrodes
2.2.2.1 Ni/Al nanorod-supported LiCoO2 cathodes
One of the most common classes of cathode active materials that are being researched and commercially used in lithium ion batteries are the lithiated transition metal oxides such as LiCoO2, LiNiO2, LiMn2O4, and the doped counterparts. Improved charge-discharge performance has been demonstrated in the literature by decreasing the average grain size of these oxide materials, as shorter diffusion length results in faster kinetics. With these practical implications, we have chosen nanostructured LiCoO2 as the active material to design positive electrodes for 3D Microbatteries. To address the objective of this work package, conformal deposition of positive electrode material onto nanostructured current collectors has been attempted.
Arrays of Ni and Al nanorods directly grown on the respective base substrates have been used as positive current collectors for the conformal deposition of nanostructured LiCoO2. Nanostructured LiCoO2 was synthesized by thermal decomposition of sol-gel precursors spray-coated onto the respective nanorod current collectors. As a first step to establish the conformal deposition process, Ni nanorod arrays have been used as positive current collectors.
The sol was prepared by dissolving stoichiometric amounts of nitrate salts (Li:Co = 1:1 molar ratio) in de-ionized water and mixing well with citric acid and ethylene glycol (1:4 molar ratio) and then evaporating at 80 deg.C for 5 h to form a transparent sol. Metal nitrates act as the cation sources, while citric acid and ethylene glycol serve as the monomers for forming the polymeric matrix. The sol was spray-coated onto the Ni nanorod arrays to cover the surface uniformly; the excess sol on the membrane surface was wiped off, followed by drying in air for 60 deg, C for 1 h. The substrate surface was carefully wiped again to remove the salts crystallized on the surface and heated at 650 ?C for 8 h in air atmosphere. The entire process was repeated 3-5 times for each sample, resulting in the formation of a thick conformal coating of nano-structured LiCoO2 onto the Ni nanorod arrays. The same procedure was followed for Al nanorod arrays to obtain Al-nanorod supported LiCoO2 deposits. Uniform and conformal deposition of active material has been achieved on both Ni and Al nanorod arrays as seen in respective SEM images.
2.2.2.2 CuSx deposition on silicon
Thin-film nanosize-particle copper sulfide cathodes have been electrodeposited on planar and 3D substrates. The morphology and composition of the cathodes has been controlled by varying the operating parameters, such as current density, pH, and temperature, of an electrolyte. The addition of a polymer to the electrolyte bath enables the formation of sulfur-rich thick porous layers. This is not possible without the additive, which serves to decrease the internal stresses in the bulk of the deposit. XPS tests showed that the CuSx cathode material obtained from the modified electrolyte is enriched in high-sulfur-content compounds (65.6% w/w). However sulfur-deficient chalcocite (36.9% w/w) is the dominating component of the high-deposition-rate samples. These samples, in addition, have the most oxidized surface (25.7% w/w CuSOx, x>1). Increasing the deposition time restores the original covellite content of the films and decreases surface oxidation (8.2% w/w CuSOx). TOFSIMS tests on CuS cathodes, showed that the lateral distribution of sulfur and oxygen-containing copper compounds is more homogeneous in modified samples compared to pristine.
2.2.2.3 Electrochemically-assisted synthesis of LiFePO4 in ionic liquids
The use of ionic liquids is most appealing for electrochemical deposition at temperatures above 100°C, and over a wider potential range than 1.2V. Electrochemically-assisted synthesis of LiFePO4 has thus been made in two different ionic liquid media: [EMIm]TFSI and [EMIm]Tf. The process, developed using anhydrous soluble FeCl3 and insoluble Li3PO4 as precursors, involved two steps: (i) the potentiostatic electrochemical reduction of Fe(III) on Pt electrode, and (ii) the chemical reaction between Fe(II) and Li3PO4 according to the following reactions:
Fe(III) + e- ? Fe(II) and Fe(II) + Li3PO4(s) ? LiFePO4(s) + 2Li+
The conditions for electrochemically-assisted synthesis of LiFePO4 were optimized by studying the effects of reduction potential, reaction time, molar ratio of two precursors and temperature. The purity of the synthesized solid had been followed by comparing the relative intensity of the most intense XRD pattern lines of Li3PO4 and LiFePO4. The molar ratio and the temperature had a huge effect on the purity of the final product. Nearly pure LiFePO4 was obtained at 275°C, after electrochemical reduction of Fe(III) at -0.3V (vs. AgCl/Ag) for 5 h, from the electrochemical bath containing Fe(III) and Li3PO4(s) with a molar ratio of 2/1.
The feasibility of synthesizing LiFePO4 in ionic liquid medium via a novel electrochemically-assisted chemical method has been demonstrated despite a limited chemical contribution, indicating that the ionic liquid acts as reducing agent of Fe(III). This is most likely due to its decomposition at 275°C, which is on the verge of its thermal stability limit. To bypass this issue, a more stable ionic liquid 1-ethyl-3-methylimidazolium trifluoromethanesulfonate ([EMIm]Tf) was used as reaction medium. Almost pure LiFePO4 has been prepared in [EMIm]Tf media containing 0.1M FeCl3 and solid Li3PO4 with a molar ratio Fe(III)/Li3PO4 of 2:1, via conducting chronoamperometry at E=-0.3V(vs. AgCl/Ag) for 5 hours at 220°C. Much higher reactivity of Li3PO4 in [EMIm]Tf leads to a lower reaction temperature, thus preserving the thermal-chemical stability of the ionic liquid solvent. Unreacted Li3PO4 and Fe3(PO4)2 were identified as impurities. Besides, higher temperatures and longer reaction times favour the formation of Fe3(PO4)2. So, within this ionic liquid medium we experienced a competition between two reactions leading, depending upon the temperature and reaction time, to the formation of multiphase materials rich in either LiFePO4:
[(Fe(II)+ Li3PO4 (s) ? LiFePO4+2Li+)]
or in Fe3(PO4)2: [6LiFePO4(s) + 3Fe(II) ? 3Fe3(PO4)2(s) + 6Li+]
2.2.2.4 Conformal deposition of MnO2
A technique to study how well films could be electrodeposited down pore structures was developed initially. Using a thin-film cell with a thickness of 1 mm, three depositions from 0.01, 0.1 and 1 M solutions of MnSO4 were attempted. The materials deposited well down this pore structure, as confirmed by profilometry data, which shows that the thickness variations increase with aspect-ratio but are of little significance at aspect-ratios up to ca. 15.
The MnO2 material deposited was found to be inactive to lithium insertion or extraction until a heat treatment was performed at 400 ?C. However, after heat treatment, a reversible capacity of around 200 mA h g-1 was achieved. A thin as-deposited planar film of this material was then tested as a lithium insertion and extraction material. A discharge profile between 4.5 and 1.5 V vs. Li was observed. A reversible capacity of ca. 50 ?Ah cm-2 was observed from a film-thickness of <1 ?m.
This material was then deposited onto carbon foam (footprint area: ~100 cm2 per cm2). A conformal deposit can be observed here with the MnO2 film clearly mirroring the structure of the foam beneath. The same conformal coating can be observed on the perforated silicon substrates. These coated microstructures have shown a capacity storage of around 2-10 mA h cm-2 depending on porosity.
2.3 Electrolytes
2.3.1 Deposition of PVdF-HFP by solvent impregnation
A hybrid co-polymer poly vinylidene fluoride – hexafluoropropylene (PVDF-HFP) has been selected for the preparation of polymer electrolyte film onto the nanostructured 3D electrode. Based on the Bellcore process, dibutyl-phthalate (DBP) plasticizer was added to the polymer to increase its liquid electrolyte uptake (and thus its ionic conductivity) and to create open porosity that will favour a rapid impregnation of the liquid electrolyte. In order to obtain a thin film with homogenous composition, a focus was made on the co-synthesis of an inorganic compound and the polymer. 3-glycidyloxypropyl trimethoxysilane (GPTMS) was selected as SiO2 precursor, to increase the polymer mechanical properties and the liquid electrolyte uptake ability. A hydrolysis step followed by poly-condensation is shown to form the inorganic network:
Hydrolysis: RSi(OCH3)3 + n H2O ? RSi(OCH3)3-n(OH)n + CH3OH
(for n= 1 – 3 and R= -(CH2)3-O-CH2-CHOCH2)
Water condensation: ?Si-OH + HO-Si?? H2O + ?Si-O-Si?
Alcohol condensation: ?Si-OCH3 + HO-Si?? CH3OH + ?Si-O-Si?
Prior to its in situ synthesis on 3D electrodes, the hybrid polymer separator was prepared on planar stainless steel electrode in order to evaluate its performance as Li-ion battery electrolyte. The hybrid polymer electrolyte film was deposited onto the 3D nanostructured Cu2Sb electrode by spray-coating technique. SEM images revealed thin polymer layer deposited onto the 3D electrode arrays.
2.3.2 Nanoporous separators for the interlaced cell design
For the 3D-IMBs, the walls of the interlaced micro-containers of the Si substrates have been made nanoporous by wet-chemical methods, and then filled with liquid or polymer electrolyte. The nanoporous separator (membrane) between the micro-containers was formed by a metal-assisted anisotropic wet-etching process. Different etching solutions, containing hydrogen peroxide, hydrofluoric acid and ethanol, were tested to obtain the proper pore-size in the inter-container partitions.
2.3.3 Electrodeposition of poly(acrylonitrile)
PAN films of controlled thickness have been made by electropolymerisation of acrylonitrile (AN) with TBAPF6 as the conducting salt, and AN itself or a 2 M solution of acrylonitrile in acetonitrile (ACN) as the solvent. Deposition of PAN onto carbon and MnO2 from solutions of AN in ACN produced electronically insulating films of uniform thickness which could be plasticized with a liquid electrolyte to give ionic conductivities from 10-5 to 1 S cm-2. A range of electrochemical techniques were used to study the factors controlling the electropolymerisation and deposition process, including cyclic voltammetry (CV), impedance and electrochemical quartz crystal microbalance (EQCM). A different approach to the deposition has also been studied, where AN itself is used as the solvent for deposition. The superoxide anion generated by oxygen reduction acts as a suitable initiator for the polymerisation, but under much milder electrochemical conditions than needed for solutions of AN in ACN.
In situ EQCM was used to monitor the deposition of PAN on Au, under CV conditions; after some nucleation steps on the first cycle, the rate of growth of the polymer was relatively consistent and hence easily controllable. Films deposited by this method have been characterised by Raman spectroscopy and optical metrology.
2.3.4 Self-assembly of surfactant oligomers as polymer electrolytes
Self-assembly of a thin (~10Å) layer of poly(etheramine) (PEA; Jeffamine®) onto LiFePO4 particles has also been shown to raise the capacity of electrode material. Since the PEA oligomer is surface active, it can be envisioned that it could also be deposited as an ultra-thin polymer electrolyte for non-uniform electrode surfaces. Thicker and solvent-free layers display conductivities of ~10-6 S/cm for conventional salt concentrations. However, for the viscous oligomer to separate the electrodes and avoid short-circuiting, the polymer layer must be strengthened mechanically. Current work with in situ cross-linking suggests that this could prove a useful strategy.
2.3.5 Electrodeposition of elastomeric network polymers
Earlier studies of the electrodeposition of PAN were extended to include the electrodeposition of elastomeric network polymers (PPEGDA) based on poly(ethylene glycol) diacrylate (PEGDA). A particular advantage of this material is that the availability of a wide range of PEGDA-type monomers potentially allows for the creation of polymer electrolytes with “tuneable” ionic conductivity and mechanical properties.
2.3.6 Self-assembly of surfactant oligomers
Work continued on self-assembly of surfactant oligomers as 3D-MB polymer electrolytes. Self-assembly of a thin (~10Å) layer of poly(etheramine) (PEA; Jeffamine®) onto LiFePO4 particles has been shown to raise the capacity of electrode material. Since the PEA oligomer is surface active, it can be envisioned that it could also be deposited as an ultra-thin polymer electrolyte for non-uniform electrode surfaces. Thicker and solvent-free layers display conductivities of ~10-6 S/cm for conventional salt concentrations. However, in order for the viscous oligomer to separate the electrodes and avoid short-circuiting, the polymer layer must be strengthened mechanically. Current work with in situ cross-linking suggests that this could prove a useful strategy. These films were then used to coat LiFePO4 positive electrodes onto Cu pillars

3. Characterisation of 3D-MB components
3.1 Current collectors
Cu and Ni nano-rod current collectors were grown using an alumina membrane in aqueous electrolytes. In contrast, the Al current collectors were formed using an ionic liquid-based electrolytic bath, either using an Al2O3 membrane or by adjusting the nature of the ionic liquid and the surfactants to adjust the morphology of the Al deposited. The grown films were characterized by X-ray diffraction (XRD) and Scanning Electron Microscopy (SEM) as described earlier.
3.2 Negative electrodes
3.2.1 Nanostructured Cu2Sb anodes
Prior to the electrochemical characterisation of the Cu2Sb electrodes, X-ray diffraction analysis was carried carried out to evaluate the content (Sb, Sb2O3, Cu2Sb) of the coating deposited onto the Cu arrays. Due to the 3D structure of the current collector and the small thickness of the active material coated onto the nano-structure, most often the X-ray diffraction peaks of the Sb compounds were masked by the strong Cu signal, resulting in low intensities. Spontaneous formation of Cu2Sb was observed after electrodeposition of Sb onto a Cu substrate. No traces of Sb2O3 were detected, whereas pure antimony was observed.
The capacity and cycling stability of the as-deposited 3D nanostructured Cu/Sb electrodes (without annealing treatment) were tested vs. lithium using coin cells. The samples were tested vs. a lithium foil. Except for the polymer separator test, the two electrodes were separated by a cellulose paper separator (Whatman GF/D) soaked with a few drops of the electrolyte. Once the coin cell is filled with all the battery components, it is sealed using a pressing tool. During the first lithiation step, an un-indentified plateau was observed at about 1.2V. Although this plateau is very similar to the reduction of Sb2O3, it is unlikely to occur here since this conversion reaction takes place at higher potentials (1.5V). It might instead be due to some parasitic reactions at the interface between electrolyte and electrode. This plateau corresponds to 3/4 of the extra-capacity (0.5 lithium) measured at the first discharge. During the subsequent de-lithiation, only 2.5 lithium were released from the structure. During this first charge and upon subsequent cycling, three plateaus corresponding to the reversible alloying of Cu2Sb into Li3Sb are clearly visible, suggesting that Cu2Sb is formed at the end of charge. The deposit exhibited a smooth homogeneous coating and showed very good cycling behaviour even at higher rates. The deposition conditions suggested for electrodeposition of Sb onto 3D Cu current collector arrays thus correspond to a current-pulsed electrodeposition process with pulses of idep=- 1mA.cm-2 for tdep= 10 ms and irest= 0 mA for trest= 50 ms.
In an attempt to improve the cyclability, the 3D electrodes were prepared by electrodeposition and annealed. The thermal annealing was used to promote the full alloying of Sb with the Cu current collectors, since although spontaneous formation of Cu2Sb was observed, traces of pure Sb were also detected as seen previously with XRD. The electrodes were annealed at 120oC in vacuum to promote the complete alloying of Sb with Cu. Independent of the amount of elemental Cu or Sb, only the Cu2Sb phase is expected to form below 300oC. The voltage profile and capacity stability of 3D electrodes annealed for 1h and 12h are plotted and compared with a non-annealed 3D Cu2Sb electrode. The voltage profiles were not affected by the annealing treatment in any of the cells tested. The typical plateaus observed during lithiation and de-lithiation of Cu2Sb are visible and the plateau observed during charge suggests that Cu2Sb is formed back at the end of the charge. The capacity retention upon cycling is greatly improved (at least doubled) by the annealing step and with the annealing time. The complete formation of the alloy Cu2Sb which presents lower volume expansion percentage than pure Sb and probably the extended availability of Cu to be re-inserted in the structure are most likely the reasons for the observed increase of the 3D electrode cycling life.
3.2.2 EPD-deposited SnO2/Cu nanorod anodes
Electrochemical properties of EPD-deposited SnO2/Cu nano-rod negative electrodes have been investigated and compared with those of electrophoretically deposited SnO2 on planar Cu foil. The as-prepared electrodes are dried with compressed air and tested vs Li using coin cells. The electrode was cycled between 0.02 V and 1.5 V vs. Li at a current density of 8?A/cm2. The initial capacity of the two electrodes is identical, indicating that they contain almost the same mass of active material. The nanostructured electrode shows much improved behaviour; during the first 100 cycles, the capacity drops similar to the planar electrode (but less sharply) and then stabilizes to an average value of 0.02 mAh/cm2 for over 500 cycles. The nanostructured electrode thus presents a better cycling stability than planar counterpart. During the initial cycles, much of the SnO2 could be a layer, filling the inter-rod space and undergoes large mechanical stress during cycling. However, with continued cycling, only the nanoparticles uniformly coated onto the Cu nanorods actively participates and thus the mechanical constraints incurred are well accommodated. Comparing the cycling behaviour of nanostructured Cu/SnO2 and Cu/Sn electrodes are compared, it is seen that the ultimate capacities of the two electrodes are identical, although the initial capacities different. The stabilized capacity of Cu/SnO2 electrode appears to result from an interfacial layer of Cu6Sn5 as in the case of Cu/Sn electrode. Excellent capacity retention has been observed for EPD-deposited SnO2/Cu nanorod electrodes, with a final constant capacity at 25 µAh.cm².
Finally, both Sb and Cu2Sb electrodeposited on Cu micro-trench negative electrodes have been investigated. Cu micro-trench current-collectors were first prepared on a glass substrate using a lithography process. A copper current-collector of height and width ca. 15µm has been obtained. Studies of the electrochemical properties of these as-formed electrodes showed no real differences compared to Cu rods coated with Sb active material.
3.3 Positive electrodes
3.3.1 Ni and Al nanorod-supported LiCoO2 cathodes
As a first step to establishing a conformal deposition process, 3D positive electrodes have been fabricated by depositing LiCoO2 onto Ni nanorod arrays. All the electrodes were fabricated with uniform thickness of LiCoO2 deposit, with a geometrical area of 1.3 cm2. Cyclic Voltammetry (CV) measurements for Ni nanorod arrays confirmed that 3D Ni nanorod current collectors are stable in the potential window from 3V up to a potential of 4.2 V vs. Li. CV experiments were performed to measure the reversibility of the nanostructured 3D cathodes by scanning from a potential of 2.5 V to an upper limit of 4.15 V vs. Li at a scan rate of 0.01 mV/s. The first two voltammograms for Ni nanorod supported LiCoO2 3D electrode were found to be identical. Two sets of current peaks (cathodic and anodic) appeared and the shapes of the curves are similar to those obtained for bulk LiCoO2. The first set is due to the first-order phase transition between two hexagonal phases, while the second peaks are caused by the order-disorder phase transition. Hence, the insertion and extraction of Li+, which is associated with order-order phase transition, dominate the charge-discharge process and provide good redox reversibility.
The charge-discharge curves for Ni nanorod-supported LiCoO2 deposits cycled at C/10 rate vs. Li displays a plateau around 3.9 V attributed to the co-existence of two hexagonal phases. The charge cut-off voltage was fixed at 4.15 V vs. Li to avoid oxidation of Ni nanorod current collector. The 3D Ni-nanorod supported LiCoO2 electrodes are thus seen to be stable on cycling vs. Li.
Electrochemical characterization have also been done for aluminium nanorod-supported LiCoO2 3D positive electrodes. The electrodes were fabricated with a uniform thickness of LiCoO2 deposit and the geometrical area was fixed to 1.3 cm2. Half-cells were cycled in a galvanostatic mode at a rate of C/10 and with a charge cut-off voltage of 4.15 V. A well defined plateau was observed around 3.9 V corresponding to the first-order phase transition between two hexagonal phases. The cycling showed negligible hysteresis and the electrodes exhibited excellent capacity retention.
Al nanorod supported LiCoO2 3D positive electrodes were also tested for their rate capability. Signature curves were collected on discharge using a cut-off potential of 3.5 V after the cells had been charged to 4.15 V at a very low current rate (C/20). For comparison, similar measurements were made on Li half-cells using LiCoO2 film deposited on a planar Al foil with 3 and 5 layers of spray coating. The electrode geometrical area was fixed to 1.3 cm2 for all the electrodes studied. Excellent rate capability is observed for the Al nanorod-supported LiCoO2 electrode compared to their planar counterparts and is shown to recover ~70% of its total capacity at a high rate of 8C. These rate capability studies thus show that nanostructured 3D cathodes exhibit high power density without compromising on their energy density - unlike their 2D counterparts.
3.3.2 Template-free LiFePO4 cathodes
A parallel effort has been devoted to the study of electrodes made by depositing LiFePO4 via a solution-casting process or electro-deposition in an ionic liquid medium via a template free process onto Al substrates. The resulting electrodes were shown to display excellent capacity retention and high rate capabilities with capacity utilization of 80% at a 5C rate.
3.3.3 Solution casting of LiFePO4 cathodes
Electrodes made of columnar and needle-like synthesized Al current-collectors were studied using LiFePO4 deposited via a solution casting process. So far the best results are obtained with electrodes constructed from columnar Al-current collectors made using ionic liquids in the presence of a CTAB additive. The electrodes showed good capacity retention and high rate capability in contrast to the case for planar aluminum, where capacity fell to 1C. Indeed, about 80% of the normalized initial capacity remains. However, taking into account the un-normalized capacity, the columnar Al-based electrode delivers a lower capacity than the flat substrate up to 1C, while exhibiting superior properties at higher rates. This is likely result from the method used for depositing the active materials. In fact, the solution casting process does not facilitate full electrical contact with all the active material.
3.3.4 MnO2 and LiFePO4 cathodes deposited carbon foam current-collectors
3D-MB positive electrodes fabricated both with MnO2 (EMD/RVC electrode) and LiFePO4 (LFP/RVC electrode) deposited on RV carbon foam current-collectors have been characterized electrochemically and with SEM.
The discharge of the MnO2-based electrode onto RVC substrates occurs over the potential range 2.5-3V vs. Li, suggesting single-phase behaviour. These 3D-MB EMD/RVC half-cells show capacities per footprint area (mAh/cm2) far greater than those of composite electrodes and 2D planar EMD/Ti electrodes. This capacity increase highlights the advantage of using 3D cathode architecture over a conventional planar configuration.
A LiFePO4 3D electrode deposited onto an RV carbon foam current-collector (LFP/RVC) has also been characterized. The LFP was deposited by spin-coating from an ink containing LiFePO4, AB, PVdF-HFP and CP. The best electrodes were obtained from inks containing 5 - 10 wt% of LFP.
A charge/discharge behaviour is obtained characteristic of LiFePO4, exhibiting a two-phase charge/ discharge plateau around 3.4 V. A significant increase in capacity is seen on going from using 5 (130 ?Ah/cm2) to 10 (870 ?Ah/cm2) wt.% inks. The rate performance of these materials is also extremely promising. The 5% foam shows a >70 % DoD (based on a full capacity measured experimentally at 0.8 C) at 14.3 C (charge- and discharge-times of 4 min.). The 10 % foam shows >65 % DoD (based on a full capacity measured experimentally at 0.2 C) at 4.5 C (charge- and discharge-times of 12 min.).
3.4 Electrolytes
A hybrid polymer electrolyte (HPE) film deposited onto the 3D nanostructured Cu2Sb electrode by spray-coating technique has been characterized electrochemically by SEM studies. Electrochemical characterization of the HPE film has been tested by assembling half-cells where the Cu2Sb 3D electrode - conformally coated with the hybrid polymer electrolyte - is tested against a Li foil counter electrode, without the use of a conventional separator. The hybrid polymer electrolyte provides pure ionic conductivity of the order of 1.10-5 S.cm-1, making it highly suitable for Li-ion battery applications. Lithium transference number in the HPE polymer separator has been measured to be 0.86 using EIS coupled with Chrono-Amperometric (CA) measurements.
Coin cells of Cu2Sb 3D electrodes covered with the hybrid polymer electrolyte soaked in 1M LiPF6 EC:DMC (1:1 volume) were tested vs. Li, without the addition of any glass fibre separator. No short circuiting of the cells was observed, confirming that the hybrid polymer electrolyte fulfils its role as separator. The cell capacity was stable for ~20 cycles before the capacity began to fade. There could be several reasons for this: the presence of water in the electrolyte structure, the loss of mechanical integrity during cycling, or the electrolyte degradation when in contact with the electrolyte or upon cycling. It was thus not possible to demonstrate that 3D electrodes of Cu2Sb coated with hybrid polymer electrolyte could be cycled without short-circuiting (without an additional separator).

4. 3D-MB assembly and characterisation
Various types of 3D half-cells with different architectures have been assembled and their performances evaluated.
4.1 Vertical-post 3D-microbatteries
Different substrates (Al, stainless steel, etc.) were tested in an electrolytic bath containing 0.1M FeCl3 and solid Li3PO4, as used for the synthesis of LiFePO4. For Al, a black deposit was observed on the Al wire in the absence of any electrochemical reaction. For stainless steel (SSA316), the solution turned green. The instability of both substrates in the electrolytic bath is due to side reactions between FeCl3 in the ionic liquid and the metallic substrates; see the schematic potential scale in an ionic-liquid medium:
Al + FeCl3 -> Fe + AlCl3 and Fe + 2FeCl3 -> 3FeCl2
Since LiFePO4 synthesis is yet to be proved successful on a metallic substrate, other organic TFSI-based salts were tested as precursors for the electrochemical synthesis. For example, the stabilities of different substrates (Pt, stainless steel A316 and Ni) were tested with Cu(TFSI)2 and Zn(TFSI)2 in an [EMIm]TFSI electrolytic bath. All were shown to be stable; the CV results at room temperature indicate the good compatibility between the substrates and metallic TFSI baths. However, it is still fair to say that the ideal TFSI-based salt for LiFePO4 synthesis has yet to be found.
4.2 Interdigitated 3D-microbatteries
4.2.1 Cu2Sb negative electrodes
Cu trenches have been used as current collectors. Cu has been deposited lithographically on a glass substrate; dimensions: 10µm thick, 12.5µm high, and 50µm separation. The electrochemical plating bath contained 0.15M Sb tartrate; the total tartrate concentration was increased to 0.45M with Na tartrate. The pH was lowered to 1 with sulfuric acid. Pulsed- and constant-current electro-depositions were performed. The deposition process has been adapted to the new current-collector designs. It was found that constant-current electrodeposition was the best method for obtaining homogeneous and conformal Sb deposits using these Cu-trench current-collectors. The general experience is that pulsed currents are to be preferred for wall thickness in excess of the trench depth. Optimization of the deposition parameters (current and deposition time) showed that Sb could be deposited electrochemically when stirring at j = -1.2mA.cm-2 for 10 min. Increasing the time and/or current density was found to lead to the formation of unwanted Sb dendrites. An annealing treatment was then performed at 120°C under vacuum for 12 h (optimal value) to promote the formation of the Cu2Sb alloy.
High-resolution SEM and electrochemical characterization was performed on the Cu2Sb electrodes. Sb electrodeposits were seen to be conformal with the Cu-trench current-collector; the thickness of the deposits (~100nm crystallites) was ~800nm. Since the Cu trenches were deposited on a non-conducting substrate, electrical contact with the Si wafers was improved using pure Cu wires. Using a Swagelock® configuration, Cu2Sb electrodes delivered a first-cycle capacity of 128µA.h/cm² at C/10 rate, and showed good capacity retention.
4.2.2 LiCoO2 positive electrodes
The chemical synthesis method used to prepare the positive material involves the use of a stoichiometric mixture of nitrate salts (Li:Co 1:1) dissolved in de-ionized water and mixed with ethylene glycol and citric acid (4:1 molar ratio). A 2-step heat treatment was used at 80°C for 24h to form a transparent sol and 650°C for 24h to form the powder.
4.3 Assembly and characterization of 3D whole-cells
A polymer separator was deposited on a negative Cu2Sb electrodes; electrical contact was established before deposition of the polymer by placing a thin copper wire on top of the electrode and fixing with conductive silver-epoxy glue. The main difficulty in assembling full cells is the problem of short circuits caused by pin-holes/breaks in the polymer film. For example, the polymer is soluble in acetone and NMP solvents, and this prevents the use of a conventional casting technique to deposit the positive LiCoO2 electrode onto the composite Cu2Sb/polymer electrode. The synthesis of the positive LiCoO2 powder directly onto the polymer was not possible because of the thermal decomposition of the polymer. Two methods are proposed to address these problems.
- The first consists of dispersing the LiCoO2 particles as a suspension in ethanol and depositing it dropwise on Cu2Sb/polymer samples. A piece of aluminium foil capable of withstanding the pressure of the full-cell between two Teflon® pieces is placed on top. The coating remains homogeneous and intact. Although the agglomerates are quite large, the positive active material fills the trenches. Because it is a freely deposited powder, some of the material was lost during manipulation of the samples.
- A second approach involved the formation of a 50µm-thick LiCoO2/AB/PVdF-HFP (80/10/10 mass %) composite film on an Al foil. This positive electrode film is then placed on top of the Cu2Sb/polymer electrode. The full-cell is placed between two Teflon® pieces and pressed. SEM micrographs show that: i) the polymer seems to be preserved after the pressing step, and ii) the positive composite film acquires the shape of the negative electrode. In this way, we can utilize the whole developed surface of the 3D cell.
4.3.1 Interlaced 3D-microbatteries
Interlaced 3D structures have been fabricated on a silicon substrate provided with arrays of interlaced micro-containers of anode and cathode materials separated by a porous-silicon wall. Etching of the interlaced structure was performed by photo-lithography and the dry reactive ion-etch (DRIE) process. The nanoporous separator (membrane) between the micro-containers was formed by a metal-assisted anisotropic wet-etching process. Li/MoOxSy thin-film microbatteries with interlaced porous silicon (PSI) membranes have been cycled. Thin-film planar cells were assembled for comparison purposes. The membrane was soaked in the LiPF6 EC:DEC electrolyte, a lithium-foil anode and a thin-film MoOxSy cathode were applied to the opposite surfaces, and the whole structure was tested electrochemically. A Li/MoOxSy battery with a commercial Celgard separator was also assembled for comparison. The sloping character of the charge/discharge curves is typical of the behaviour following insertion of lithium into a single-phase host material, like MoOxSy. In agreement with conductivity data, the charge/discharge overpotential is higher than that for the cell with a commercial Celgard separator. This is reflected in the lower capacity of the cell with PS.
It is encouraging that both cells showed stable reversible cycle life with a capacity loss of about 4%/cycle. Finally, the 3D interlaced nanoporous silicon substrate was filled with the hybrid-polymer electrolyte membrane, the graphite-anode and LiCoO2- or LiFePO4-cathode material. Insertion of the electrode materials was carried out by several successive centrifugation steps. Assembly and testing of the full 3D-IMBs continues.
4.3.2 Aperiodic 3D-microbatteries
Conventional spin-coating techniques have existed for many years and the literature is replete with examples of this process. A novel economical approach has here been developed in which a computer fan was mounted in a Perspex box, an alligator clip attached perpendicular to the centre of the fan with a quick-setting epoxy glue. The carbon-foam substrates were then prepared by attaching a small titanium-foil tab with a quick-setting epoxy resin. The foams were then dipped into the ink for a few seconds to allow penetration into the structure. Filling was achieved by capillary action. The Ti tab was then secured in the alligator clip. By the application of a fixed voltage, rotation rates were calibrated to be in the region of 2000-3000 rpm. The rotation was then applied for 30 s and the sample then removed. Further dipping and spinning steps can then be applied if a higher degree of cathode loading is required on the foam struts.
Solutions of a mixture of diacrylate and monoacrylate polyethylene glycol-based monomers in 0.5M LiTFSI in propylene carbonate were used to electrodeposit polymer electrolyte on electrode surfaces. The diacrylate monomers were either polyethylene glycol diacrylate (PEGDA, Mn~700) or SiO2 nanoparticles (~20nm) functionalised with methacrylate groups. The particles are suspended in the diacrylate monomer in a 10%(w/w) ratio. The monoacrylate component is polyethylene glycol monomethylether acrylate (PEGMEA, Mn ~480). The deposition solution is a mixture of 15%(v/v) diacrylate and 4.5%(v/v) monoacrylate in propylene carbonate with 0.5M LiTFSI added as electrolyte. The polymer electrolytes are deposited from this solution under cyclic-voltammetry conditions over at least 10 cycles at 100mV.s-1.
The ultimate challenge of the project has been the building and testing of a full 3D cell. The following cell design has been developed as a reliable means of construction. A polypropylene vial has a nut, heat-sealed into one side of the tube. The carbon foam with the primary electrode layer is then placed in the tube and a bolt is used to make compression contact to the foam as well as acting as a current terminal. The setup is then used to electrodeposit an electrolyte on the primary electrode and contact made with the nut and bolt assembly. If a sedimentation method is used to provide the electrolyte coating, the vial is simply filled to the desired level with a solution containing the dissolved electrolyte. The excess solvent is then removed to leave a conformal coating. Once this coating has been applied, the second electrode can be added by sedimentation or other techniques and a contact can be made.
The coverage after spin-coating with ink was investigated as a function of the wt % of solids in the ink. Four compositions were chosen: 5, 10, 15 and 20 wt%). The resulting coverage of the cathode material after spin-coating is shown in Fig. 3.2. The coverage is reasonably continuous over the surface of the foams. In the 5 and 10 wt.% inks, the coverage is patchy in some areas, making it difficult to quantify the thickness of the coverage, but it is clear that it increases with the ink wt %. The coverage resulting from the use of 5 and 10 wt% inks is typical of the whole surface of a 0.5cm2 electrode. Examination of the cross-sections of these samples showed consistent coverage throughout. No significant pore blocking was observed in these foam structures. The image of the foam prepared from the 15wt.% ink shows a rather thick deposit on all the struts. However, significant pore-blocking could be seen similar to that seen clearly in the 20 wt.% sample. It was concluded that the 5 and 10wt.% inks resulted in the best coverage and were the most suitable preparative methods for further testing.
The performance of the 5 and 10 wt.% samples both show the typical plateau at about 3.4V vs. Li for LiFePO4. The specific footprint capacity for the 10wt.% ink is roughly twice that for the 5wt.% ink, indicating that increasing the wt % of the ink results in a direct increase in the amount of material deposited on the foam. Thus, the highest capacity that can be achieved with this approach is ~320µA.h.cm-2 for the 10 wt.% ink. A higher ink wt % will result in a further increase in capacity. However, pore-blocking begins to make these electrodes unfeasible as hosts for a full 3D-microbattery construction. Initial results showed the desired coverage and reasonably competitive footprint-area capacities. However, further improvement is required. To this end, attempts were made to coat the substrate with several layers of deposit to enhance the capacity while maintaining a structure with no pore-blocking. The SEM images of the deposits on the foam after 1, 2, 3 and 4 spins are difficult to distinguish for one another. However, the coverage is uniform and no pore-blocking is observed. To quantify the coverage remaining after the preparation of the substrates, the foams were cycled galvanostatically in a bath of electrolyte. The loading is seen to increase almost linearly with each coating. The capacities obtained here of 1500µA.h.cm-2 are extremely high and the structure remains porous and open with room to accommodate the subsequent electrolyte, anode and current-collector layers.
The rate performance of the materials is satisfactory at higher rates with the three-coat electrode achieving 50% discharge capacity at 25C - the normal value for this type of electrode is only ~10C in a planar set-up.
Polymer electrolytes were electrodeposited under cyclic-voltammetry conditions on LiFePO4, which itself was on 100ppi RVC substrates. The polymer electrolyte is easily identifiable by the wrinkled texture as a result of SEM electron-beam damage; this has also been confirmed by EDX. The images show complete coverage of the LiFePO4 coating in all cases. However, pore-blocking is seen when thicker films are deposited.
Early efforts to fabricate a full 3D microbattery were made using foams coated by successive layers of electrode and electrolyte. An TiO2 ink was added to the polypropylene vials, and forced to the base of the tube by centrifugation. This was then tested as a full cell and initial results showed a stable OCV.
X-ray tomography images were made of foam coatings. A polymer electrolyte with LiI instead of LiTFSI as the salt was here used as the TiO2 coated composite electrode. Single 2D slices can also be created. As an example, in the higher-density LiFePO4 is seen as a bright coating around the darker RVC substrate. The coating is quite uneven in places throughout the foam, but porosity is largely retained. In the case of sedimentation of the second electrode by centrifuging, CT reveals that the coating extends throughout the foam. However, the centrifuging process does not fill the pores, but instead, material tends to build up around the outside of the foam. Stacks of foam with various coatings can be imaged simultaneously. The coatings are seen to be conformal and of even thickness throughout the depth of the foam.
4.3.3 Concentric 3D-microbatteries
Concentric 3D half-cells have been assembled on perforated-silicon substrates with copper sulfide (CuS) and lithium iron phosphate, LiFePO4 (LFP) cathodes and duly tested. The experimental procedure for the preparation of the CuS cathode by electrodeposition, characterization of the structure and composition of the cathode material, and assembling of the cells has been described earlier.
An electrophoretic deposition method has been used for the first time to prepare thin-film 3D-LiFePO4 (LFP) cathodes. The effect of polymers and surface-active additives in the electrolytic bath, voltage and deposition protocol have been studied with the aim of obtaining highly-adhesive LiFePO4 films to be utilized in 3D microbatteries. The samples were investigated by scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDX), XPS, TGA and TOFSIMS. The results of these methods confirmed the presence of polymer binder and its homogeneous lateral distribution in the composite EPD-LiFePO4 cathode. It was found that the higher the voltage used for EPD, the larger the number of aggregates formed in the EPD-LFP film. Modification of the electrolyte by PVdF eliminated cracks and peeling of the EPD films and enabled preparation of 5-30 micron-thick cathode films highly adherent to the substrate. However, a few large particles still appeared in the EPD cathode film deposited at 100V. In contrast, relatively uniform deposits with smooth surfaces were obtained from well-dispersed and stable LFP suspensions containing Triton X-100 (TTX) as dispersant and Black Pearl carbon as the agent used for improvement of electron conductivity of the LFP cathode. TGA showed that the EPD-LFP cathode contains 1.5% PVdF. The incorporation of additives is seen to decrease the polarization. The pristine-cell exhibited 75mV charge/discharge overpotential, a value more than three-times larger than the overpotential observed for the modified-LFP cell. Additionally, the discharge profile of the modified-LFP cell follows a well-defined voltage discharge plateau at 3.3V, while a sloping character is observed for the pristine-LFP cell. The discharge capacity of the modified cathode (163mAhg-1) is very close to the theoretical value for LiFePO4 of 170mAh g-1. The cycling performance of the 3D half-cell comprising modified EPD-LFP was evaluated by cycling at 0.2C or 0.28mA/cm2(battery footprint) at room temperature. The discharge capacities were stable for 100 cycles with a capacity loss rate of 0.01% per cycle. The planar cell with the modified cathode exhibited about 100µAh/cm2 capacity at 0.08C, while the modified semi-3D LFP cell showed about 1.6mAh/cm2(battery footprint) reversible capacity. This agreed with the geometrical area-gain factor (AG) of 9 for the perforated-Si substrate. Concentric 3D Li/LiFePO4 half-cell microbatteries on perforated silicon substrates showed a peak-pulse power capability of 175mW/cm2.
3D-MB half-cells using a perforated Si substrate and CuS as cathode material have subsequently been tested electrochemically. This type of half-cell can be cycled at constant discharge current for more than 400 reversible cycles with ~0.1%/cycle capacity loss. The range of current densities used both for charge and discharge varies from 100 to 2000 µA/cm2. In the latter case, the capacity drops to ~50% of its value measured at low current densities. The pulse power capability of this system is high enough. For discharge pulse-steps of 100 ms, the cell provides ~ 160 mW/cm2 pulse power. At a discharge rate of 30 mA/cm2, the CuS cell delivers 2030 30 ms-long pulses with a capacity of 0.54 mAh/cm2 and an average voltage of 1.7V.
3D-MB half-cells using an Au-coated perforated Si substrate and LiFePO4 as cathode material have also been prepared using electrophoretic deposition. Electrochemical characterization and cycling tests have been performed and the results compare both to planar and non-modified electrodes. After 20 cycles, the LFP-based modified electrode provides a capacity of 80 µAh/cm2 with an average potential of 3.4 V. The pulsed-power capability tests gave a pulse-power peak of ~ 175 mW/cm2 for a 1s discharge pulse-step. The pulse-power peak decreased to ~ 58 mW/cm2 for a 10s discharge followed by a 5 min rest period.
4.3.4 Summary of 3D-MB assembly and characterisation
A. Two methods of assembling interdigitated “trench-type” 3D-MB have been used:
- dropwise insertion of LiCO2 into the trenches;
- attachment of the composite cathode to the anode/polymer structure the application of
gentle pressure.
B. Spin-coating techniques have been developed to deposit LFP cathode material onto high-surface area carbon foams. LiFePO4-based 3D-MB half-cells have been assembled successfully with aperiodic architecture giving a capacity of 1.5mAh/cm2. Complete 3D-MB cells with components TiO2/solid polymer electrolyte/LFP with aperiodic architecture, exhibited a stable OCV.
C. Electrophoretic deposition procedures have been developed for the preparation of 3D-LFP cathodes, 3D-composite ZrO2-PEI ceramic membranes and 3D-MCMB-based anodes. Assembled 3D-MB concentric Li/LFP half-cells gave a 1.5-2.5 mAh/cm2 capacity >100 cycles – and >200mW/cm2 pulse-power capability. Concentric Li/3D-MCMB half-cells with an EPD-anode showed a reversible capacity of ~370 mAh/g.

Although electrochemical testing of assembled 3D-MB full-cells is still underway, the power- and energy-density characteristics of our 3D-MB half-cells are already out-performing the best commercial thin-film planar batteries available today.

5. 3D-MB device packaging and characterisation
5.1 Packaging solutions tested
From a variety of different packaging solutions, three different test formats have been chosen for the project. The flexible flat cell LFP has been used for very thin samples with large footprint-to-height ratio. e.g., for 3D-structured silicon wafer substrates. While MC 614 and V500 cells were used for foam-based electrodes with larger sample height.
5.1.1 A flexible flat cell
The packaging solution for very thin electrodes - below 300 µm thickness - has the flat-cell housing of a flat cell. The cell basically consists of two copper foils as positive and negative current-collectors, which are linked together by a polymeric sealing. The overall thickness of the cell is less than 500 µm.
5.1.2 A 614-type coin-cell
For cathode materials with potentials above 3V, a combination of copper and aluminium foils has been established for negative and positive electrodes, respectively. An advantage of this unique cell type is the possibility of integrating the flexible metal housing directly into MEMS devices. Trials with silicon substrates have been conducted. The dimensions of the gold-plated 3D-silicon substrates are 10 x 10 mm² with a thickness of 500 µm. The direct electro¬deposition of active material onto the current collectors/housing has been tested. These showed that care must be taken not to damage the housing with the sample during the sealing-process and vv. Parameters which can be adjusted are: the substrate dimensions (area, thickness) as well as the conditions during vacuum sealing.
Another option of a MEMs-compatible housing is a tiny coin-cell, the MC614. With a diameter of 6.8 mm and a height of 1.3 mm, this cell can be integrated into very small device. Each cell consists of 3D-coated carbon foam (Reticulated Vitreous Carbon, RVC) as cathode and a lithium anode. MnO2 and LiFePO4 have been used as active cathode materials. In the case of MnO2, the samples were prepared by electrodeposition; LiFePO4 was made using the spin-coating technique .
The capacity of the coated foams depends almost entirely upon the thickness of the deposition (µAh/g). The surface area of carbon foams is about 2.5 cm2 (for a 4.5 mm x 0.8 mm sample). Two types of foam electrodes have been incorporated into small 614-type button cells:
- MnO2 has been electrodeposited on compressed carbon foams with 80ppi. The nominal capacity of the cathode is ca. 120 µAh.
- LiFePO4 samples have been prepared on 100ppi foams by spin-coating; the samples are provide are ca. 50 µAh.
The lack of robustness of the carbon foams is a critical issue. Especially during the cell closing process – crimping of the top and bottom casing – the carbon material was is brittle and easily damaged, leading to a large number of defective cells.
5.1.3 A V500-type coin-cell
An alternative approach for a 3D microbattery has used a cell-housing with a larger volume and footprint area. With this design, a stack of 3D-foam electrodes can be assembled leading to an interdigitated semi-3D full-cell. With this packaging option, both anode and cathode are composed of 3D-structured Ni-current collectors coated with active material. The thickness of each foam electrode is in the range of 500 – 800 µm, which exceeds traditional electrodes by a factor of 5 – 10 for a given footprint. Electrodes have been tested for this cell type, which allows the combination of different anode and cathode materials - resulting in adjustable capacities and cell voltages.
5.2 Electrochemical characterisation of prototype batteries
5.2.1 Characterisation of 614-type coin cells
The cathode materials MnO2 and LiFePO4 have been tested in 614-type button cells. The cells were assembled with the 3D-coated carbon foam cathodes, a glass fiber separator soaked with liquid LiPF6-based electrolyte and a lithium metal anode. The geometrical area of the cathodes was 0.17 cm2 with a specified cell capacity of ca. 120 µAh for MnO2 and 50 µAh for LiFePO4 samples. Only a small number of the assembled cells turned out to be usable. Either a drop in cell voltage or an electrical short circuit has been observed in the others. Cell assembly also revealed handling problems of the carbon foam electrodes. Different failure modes were observed:
- A time dependent drop in cell voltage, associated with an increase of internal impedance (decomposition, leakage, contact loss?)
- Internal short circuiting after assembly; mechanical stability seems insufficient for the processing small coin-cells.
Although the RVC carbon foam substrates are relatively mechanically robust and can withstand a reasonable amount of pressure, they cannot be compressed. During crimping of the top and bottom parts of the housing, the force applied on the foams appears to be too high in most of the trials. Apparently, the foams or broken off particles can penetrate the separator, resulting in a short- circuit to the lithium anode.
Electrochemical cycling performance of LiFePO4 cells showed the initial discharge capacity to be 39 µAh (312 µAh/cm2), which is in the range of the specified capacity of ~50 µAh. Initially, the cell showed a good cycling stability (96.5% remaining capacity after 39 cycles), but after 39 cycles the capacity dropped to zero - most likely a result of an internal short-circuit of the cell.
Electrochemical cycling performance of an MnO2 electrode showed an initial discharge capacity was 99 µAh (corresponding to 791 µAh/cm2), which is lower than the specified range of 120 µAh g. The capacity retention of the cell is rather poor during the first cycles, and drops to zero after cycle 35. Again, an internal short-circuit can be suspected. The cells had lost 46% of their initial capacity after 35 cycles.
5.2.2 Characterisation of V500 3D full-cells
Electrochemical evaluation of V500 cells has not been possible. The cells exhibit too severe degradation phenomena after the initial cycles. It would seem that there is an intrinsic instability in the system related to the use of nickel-foam in a lithium-ion cell at potentials above 3V. Internal short circuits have also been observed after a few cycles.
Initial cycles of a LFP vs. TiO2 cells, where both electrodes are coated on nickel foam substrates and separated by a 30 µm separator, showed failure after only two cycles due to an internal short circuit.
A first attempt to charge an NMC-cathode cell showed that side reactions occur above 3.5V, which lead to a decrease in cell potential during constant current charging. The cut-off voltage of 4.2V could not be reached for either anode-cathode combination NMC/C or NMC/Si-C.
For LiFePO4 (LFP), the upper cut-off potential was limited to 3.4V. Initial cycles of LVF vs. hard carbon showed that the cell potential drops during constant-current charging due to side reactions after a couple of cycles. A short circuit occurs during cell discharge of an LFP vs. Si-C cell. Up to this point, it has not been possible to get the system to work for a reasonable number of cycles. The reversible capacity is also insufficient for all anode-cathode systems being tested.
5.2.3 Characterisation of 3D-CuS/Li cells
Semi-3D half-cells were also tested. The batteries were of coin-cell type 2032, comprising Li-metal anodes, a Celgard separator soaked in electrolyte (LiPF6 in EC:DEC 1:1 + 2% VC) and 3D-CuS cathodes. The geometrical area of the cathode was 0.49 cm2 and the specified cell capacity was in the range 450 750 µAh. The initial discharge capacity was lower than the specified range of 450 750 µAh. During performance testing for rate capability up to 3C, a comparable performance has been observed with respect to the cell capacities. For higher rates, however, the values for different sample cells differed significantly. One of the cells still delivered 6 – 7% of its nominal capacity (at C0.5C) at an impressive rate of 100C, corresponding to a discharge current density of about 5000 µA/cm². On the other hand, about 10% of C0.5C has been obtained for another cell at 20C. All cells exhibited slight capacity fading with increasing cycle number. Constant current/constant voltage (CCCV) cycling showed one of the cells to be capable of very fast 10C/10C charging/discharging, though only 25% of the initial capacity remained after 1000 cycles. This same cell could be cycled at 5C/5C rate for 250 cycles with 54% of the initial capacity remaining.
Rate capability was evaluated for a second batch of semi-3D CuS/Li microbatteries. The cells were cycled in the range 1.90 - 2.45V at low C-rates, because of the irreversible processes that occur below 1.8V. However, at high C-rates, the voltage cut-off at discharge has been lowered to 1.8 and 1.6V, respectively, dependent on the rate. The temperature dependent rate performance up to 20C has been evaluated in the sequence 20°C, 30°C, 40°C, 50°C, 0°C and 10°C, and has been repeated for 20°C. Strongly differing cell capacities have also been observed in this second batch. From an initial CC-CV charging between 1.9-2.45V (assuming 0.45 mAh at 0.1C rate), capacities of 1.5 mAh, 0.9 mAh and ca. 0.25 mAh were obtained for three different cells. Strong capacity fading was clear for all cells, judging from the decrease of capacity at constant load. The actual rate performance strongly depends on the initial capacity of the cells, hence the different film thicknesses. Discharge of 20-30 % of initial capacity has been obtained at 20C. Cycling stability of the second batch was thus inferior to that for the first batch. Comparison between different temperatures is not possible due to the strong continuous degradation during cycling. Cell capacity had dropped dramatically after resting for two weeks before the low temperature testing. Possible reasons for this behavior are cell leakage and/or electrolyte degradation.
5.3 Proof-of-concept
A major goal is to develop efficient power sources for small and micro-sized electronic devices in the field of medical implants, such as pacemakers, neural simulators and drug-delivery systems, and in small autonomous devices. The demonstrator systems have to be confined within an appropriately durable packaging for industrial cell-testing procedures. The promising candidates suggested by earlier were tested to achieve a “proof-of-concept” and validate the cells as realistic power sources in prospective micro-devices. The analysis of the demonstrators was carried out at “body temperature” (37°C) to reproduce realistic conditions for implantable or near-body applications in medical devices.
5.3.1 CuS semi-3D microbatteries
The 3D half-cell test cells were of coin-cell type and comprised a Li-metal anode, Celgard separator soaked in electrolyte (LiPF6 in EC:DEC 1:1+ 2% VC) and a CuS cathode. The geometrical cathode area of was 0.49 cm2 and the specified cell capacity was in the range 450 750 µAh. The electrochemical testing included constant-current and pulse testing under discharge.
5.3.1.1 Constant-current testing
The test cell showed a charge/discharge overpotential of ~138 mV on the 5th cycle. This is a signature of a rather high internal cell resistance that would lead to energy losses in the charge/discharge cycles. The initial discharge capacity was between 300 350 µAh (corresponding to 600-700 µAh/cm2), which is lower than the specified range of 450 750 µAh. The cells showed a capacity loss of 2-15 % per cycle. This is significantly higher than the 0.09-0.15% capacity loss per cycle reported earlier for testing performed at room temperature. The cells had lost 306 µAh after 200 cycles compared to the initial capacity, corresponding to a total capacity loss of 88 %. It can also be noted that part of the charge capacity is irreversible, which can be attributed to side reactions occurring during charge, such as oxidation of the electrolyte.
5.3.1.2 Pulse-testing under discharge
The discharge capacity for cells discharged under pulsing was compared to constant-current discharge. The initial delivered discharge capacity was 455 µAh (corresponding to 928 µAh/cm2). The loss in discharge capacity for the pulsed cell is 341 µAh over 100 cycles which is 15-40% more compared to cells cycled at constant current. This may indicate that pulsing of the CuS cell has a negative impact on the cycling performance. The minimum potential decreases for higher pulse currents and with increasing cycle number. There is no significant difference in pulse shape for pulse #10 and #100.
5.3.2 MnO2 semi-3D microbatteries
3D half-cells were assembled as coin-cells with an MnO2 cathode, a Li anode, with the separator soaked in electrolyte. The geometrical area of the cathode was 0.17 cm2 and the cell capacity was ~120 µAh. Electrochemical testing included constant-current and pulse testing under discharge.
5.3.2.1 Constant-current testing
The cells were cycled at 37°C and showed a charge/discharge overpotential of ~592 mV on the 5th cycle. This is a signature of a very high internal resistance in the cell, leading to energy losses in the charge/discharge cycles. It can also be noted that ~200 µAh/cm2 of the charging capacity is irreversible, which could be attributed to side reactions occurring during charge, such as oxidation of electrolyte. The initial discharge capacity was in the range 101-132 µAh, which is in rather good agreement with the specified capacity of ~120 µAh. The cells showed a capacity loss of 1-9 % per cycle. After 25 cycles, the total capacity loss was 14%, 78% and 48% for the three test cells, respectively, and after 70 cycles 52% and 85%, respectively, thus implying a large variation in performance between cells.
5.3.2.2 Pulse-testing under discharge
The discharge capacity for the cells discharged at constant current with pulsing during discharge in comparison to pure constant current discharged cells indicated that pulsing of the cells had a negative effect on cell performance, in that the delivered discharge capacity was lower for pulsed cells compared to cells discharged at constant current. However, considering the variation in performance of the cells tested under constant-current discharge, these differences may well be within the cell-to-cell variation for this type of laboratory cell, rather than reflecting an effect of the cell-test mode used. The pulse shape of pulse #1, #5, #10 and #70 (corresponding to pulse currents of 0.1, 0.5, 1.0 and 10 mA, respectively) suggest that the minimum potential decreases on increasing the pulse current. The slope of the voltage curve directly after the pulse is flatter (voltage relaxation is slower) for pulses with higher pulse currents applied. This implies that the lithium transport and rearrangement within the MnO2 material to reach equilibrium after a high current pulse is slow and that the materials rate capability is rather low. A decrease in minimum potential and a slower voltage relaxation are observed with increasing number of cycles when the pulse current is kept the same. This indicates that these MnO2 cells are negatively impacted in terms of performance even for rather moderate pulse conditions (1 pulse/cycle; 1s pulse duration; ?6 mA/cm2 pulse current).
5.4 Implementation in normal-scale Li-ion battery production.
An extra bonus of the project has been the feasibility of incorporating 3D-MB concepts into conventional Li-ion battery production; this has been evaluated here. The most interesting candidate was the aperiodic “sponge” architecture, where the solid network of the “sponge” serves as the substrate and electronic conductor for lithium intercalating anode and cathode materials. The sponge or foam material can, in principle, be adopted in a roll-to-roll coating process. Most suitable for this process is a metallic material like Ni-foam due to its mechanical properties. Based on this idea, a design study has been conducted. Different types of 3D-coated Ni-foam electrodes have been incorporated in Li-ion cells comprising an interdigitated electrode stack, where alternating anodes and cathodes are separated by thin membranes to prevent short circuiting.
The translation of this cell design into a production process was hampered by the electrochemical stability limits of the Ni-foams. Side reactions (corrosion) at the positive electrode occurred at higher potentials during charging of the battery, which led to sudden breakdown of the cell. This intrinsic problem has yet to be solved. Other materials or material combinations will be needed to make the system function more satisfactorily electrochemically, for subsequent implementation in production technology. Financial aspects must also be carefully evaluated.

By the end of the project, “proof-of-concept” for 3D-MBs had been achieved at a design level, but no pilot-line production of 3D-structured electrode materials and cells was yet in function.

Potential Impact:
It is important to be aware that this particular project places special demands on the competence of the partners involved; no single country, in Europe or elsewhere, has access to the accumulated competence required to derive optimal gains from such a project. A multinational project is essential to make any meaningful advances. A significant percentage of the World’s entire documented competence in the area of 3D-MB has been assembled in a concerted effort within SUPERLION.

Strategic impact:
Technically, the research here targets significant innovations in performance, cost effectiveness, and the electrical quality of miniature power systems, through advances in materials, processing, device structures and systems integration. The expected significance of the research has both scientific/technological and commercial aspects: first; research on new, rather complicated materials, and the development of methods for their deposition as conformal thin-films and 3D-structures can lead to high-capacity batteries on a variety of substrates, rigid and/or elastic; second; a wide range of applications is awaiting the development of microbatteries: smart cards and anti-theft chips, sensors, miniature RF transmitters, microrobots, biochips, implantable medical devices, MEMS devices, etc. The development of a generic “enabling technology” to a pre-commercial level is very attractive for Europe, particularly in view of the existing strengths in the areas of micro-electronics fabrication, European battery companies, on-going research in MEMS-platform technologies – and, not least, in the medical micro-implant industry. The proposed thin-film and 3D-MB technology, although requiring extensive research, is conceptually relatively straightforward compared to the state-of-the-art microelectronics technology. The following technologies are the main potential long-term outputs from this project:
• Novel 3D-MBs for MEMS and medical applications.
• Novel multi-functional nano-materials for electrodes and electrolytes.
• Novel technologies and processes for MB fabrication.
It is anticipated by the electronics industry that our daily lives will be changed markedly by the advent of autonomous devices in Medical Systems and Ambient Intelligence. Clearly, access to small-size batteries (3D-MBs) to power such devices will be an important factor in their development.
Military applications involve an intricate network of sophisticated electronics systems enabling communication, information exchange, situation awareness, and other advanced capabilities. All these drive the military’s need for ever-increasing amounts of efficient and reliable micro-power sources; especially for remote sensors; friend or foe identification; embedded sensors for system integrity monitoring; communications systems monitoring, typically of satellites; low-power mobile displays; flexible sensing surfaces; etc. One analysis of MEMS sensor and accelerometer applications is that almost 100% of the sensor market, and roughly 30% of the accelerometer market, would use microscopic batteries if they were available. On this basis, it is estimated that a market for microscopic batteries in these two fields would be $50M/year, if such a product existed. Other significant markets would also develop if microscopic batteries could be provided such that, within a decade, microscopic batteries would have a market of over $100M/year. A requirement for mobility excludes standard wired power sources. A requirement for autonomy excludes primary battery systems that cannot provide power to integrated systems for extended periods. Requirements for small-size, extensive integration and large numbers of units exclude the use of coin-type or standard format batteries because of the difficulty of mounting such batteries into the format required by integrated systems. Microscopic batteries, once available, will have performance advantages that will prove to be critical to specific system applications, such as multiple, definable voltage levels, lower power requirements, and better power distribution.
The major obstacle in using batteries in MEMS is the size and weight of available batteries. To date, large external batteries have been used. Internal batteries must be microscopic. Dimensions must be in micrometers, rather than centimeters, and good specific power and specific energy must be available. Presently, the smallest external batteries available commercially are of the order of 0.1 to 1 cm3 in volume and 1 to 3 g in weight, and the greatest difficulty in their use to power MEMS, besides their size and weight, is the fact that all such batteries are primary and are not secondary or rechargeable batteries. Rechargeability is mandatory in many MEMS applications. Batteries for internal MEMS applications would need to have several important characteristics. First, many MEMS applications require the capability of large numbers of repeated charge/discharge cycles. Second, they must have a minimum internal resistance to limit energy losses during battery operation. Third, they must be robust, so that changes in temperature, pressure, and other conditions do not damage performance. Fourth, MEMS batteries must be produced in large quantities, at low cost, and low rejection rate. What is required is an entirely new class of micro-batteries with peak specific powers much higher than present batteries, with specific energies many times that of capacitors, and which are built on a microscopic scale, suitable for internal integration either into an existing MEMS, for retrofit purposes, or unitarily fabricated as part of the MEMS, for original manufacturing purposes.

Economic impact:
The MEMS market has recently (in 2010) been estimated to be worth an incredible 9.7 billion USD; of this, 8-10% will be realized by MEMS-related industries and contract manufacturers. More than half of today’s Systems companies who have integrated MEMS factories will be using external manufacturers. Numerous MEMS spin-off companies will be created by large integrated companies. IC manufacturers will be more and more involved in MEMS manufacturing. Today, a large fraction of MEMS production presently occurs in two areas, gyroscopes and accelerometers, which markets have increased by 55% between 2005 and 2006. In 2006, the market reached 232 MUS$. An average 35% CAGR for 2006-2011 periods is expected. Other types of MEMS applications are emerging rapidly: the microfluidics market for research applications will steadily grow in the coming 5 years, but the double digit growth will come from the diagnostic market. Indeed microfluidics has a strong competitive advantage in this market enabling analysis automation and decentralization, but also in facilitating the molecular diagnostic emergence and success. Microfluidics components for diagnostic are expected to reach €1B market value in 2011 with a total accessible Market of about € 5B.
The broader impact of this proposal is its effect on the participation of today’s youth in the fields of Science and Engineering. The work plan involves European graduate students in research programs Uppsala, Amiens, Toulouse, Tel Aviv and Southampton Universities. It can be stressed that this area of micro-battery R&D is quickly acquiring a worldwide market, with a strong competition between international groups.
Successful implementation of the project will thus allow the participating institutions to strengthen their positions as strong players in MEMS-related fields. This applies especially for St Jude Medical and VARTA. New 3D-MB batteries will not only provide cost-reduction for existing autonomous devices, but also will improve basic performance of these products. One may expect profitability of the MEMS market to be high, possibly comparable with that of the semiconductor industry. Important potential benefits on employment can also be expected.

Social impact:
On a general plane, every moment we spend at home (or away from work) should be enjoyable and uncomplicated: time spent “recharging our batteries(!)” or simply enjoying ourselves. This is why Philips continue to develop products designed to help make our life simpler and more relaxing, whether we are watching TV, adjusting the lighting in our bedroom, cleaning our teeth, listening to music or simply making our coffee in the morning. However, our everyday environment is also inseparably dominated by our state of health; to this extent, any effort made to improve our standard of life is important, not least for the many amongst us suffering from heart disease. The European Society of Cardiology estimated in 2000 that cardiovascular disease is the cause of four million deaths a year in Europe. Mortality from the disease is rising in Europe. Predictions state that the disease will be responsible for 19 million deaths annually worldwide by 2020. Cardiovascular disease also places a heavy burden on Europe's economies. It has been estimated that 74 billion Euros are spent annually by EU governments treating cardiovascular disease. Another 106 billion Euros are incurred in extra costs stemming from reduced productivity and the economic impact of illness and death. For these, all improvements in the quality and reliability of their “life supporting” pacemaker (or similar) will be a most welcome advance. Thus, implementation of the project will lead, in a medium-term perspective, to improvements in the quality of healthcare, which is becoming increasingly important as the average of the European population is increasing.
Medical applications always tend to stretch the ingenuity of the battery designer to the limit. Total mechanical integrity and reliable lifetime characteristics are of interest for most battery systems. However, these are absolutely vital for all medical applications and life-critical for implanted systems. Significant efforts have been made in recent few years to provide compact and lightweight batteries to make possible many detection and treatment methods that were not available only a few years ago; but much progress needed. Availability of new 3D-MBs will spearhead the development of new medical and daily-life technologies and create new markets, where the participating European institutions will play leading positions.
Implantable electronic technology is not limited to cardiac conditions. Therapies like drug delivery, pain management, treatment of neurological disorders such as Parkinson's disease and many others are also available or emerging. These products not only simplify the testing, monitoring and treatment processes, but can also help to improve the quality of life for the patient by minimizing time spent in hospitals and often providing automatic, continuous treatment of chronic conditions. As Medicine gains a greater understanding of disease mechanisms, devices targeting new patient groups are being developed and new patient groups are benefiting from implantable devices. Technological advances in implantable medical electronic devices provide increasing number of therapies requiring additional power supplies. At the same time, decreased device volumes are advantageous both for patient comfort and for access to various part of the body. Greater functionality and greater miniaturization are driving the demand for new power sources for implantable devices in the future. .
This joint EU project has facilitated the further development of new battery technologies that are significantly smaller and more powerful than what is state-of-the-art today. This will potentially have a large impact on future implantable medical devices.

Environmental impact:
The emergence of micro-batteries to power a range of autonomous electronic devices will have two major environmental impacts: such autonomous sensing devices will help to decrease the energy consumption in our living and working environment. At the same time, the levels of pollution will decrease - it had been estimated that 25% of all CO2 emissions is produced from energy used to create a comfortable living environment (at home or at work). The use of autonomous sensors will also facilitate environmental monitoring, which helps to address environmental problems.
The project thus involves the dual features: future energy storage systems and reduced environmental hazards (less toxic cathode materials, lower energy-consumption processes, etc.). In this sense, it can also be seen as contributing to the Building of a Sustainable Society.
A 3D-MB design strategy taking advantage of the impressive intrinsic gains of moving into the third dimension has thus emerged in recent years as an exciting new direction in electrochemical energy storage systems. Much is said about the desperate need for new battery materials to improve battery performance, while the possibilities of improving battery performance through moving their electrode geometries into the third dimension are largely neglected. An important paradigm shift is well underway, however, and one which has been accelerated significantly by the concerted efforts of SUPERLION. A vital feature of this type of 3D-MB design is that it provides a strategy that allows us to break the classic compromise which must normally be reached between energy density and power density. In 3D-MB batteries, we achieve BOTH through the use of nano-materials and nano-technologies! However, present 3D-MB approaches still rely on costly micro-lithographic techniques, resulting in an expensive battery which still cannot meet full market requirements. Nano-architecturing approaches involving the use of new processes - such as assisted template synthesis, electrodeposition, electrophoresis, infiltration and impregnation techniques - are now available, however, at relatively low cost. SUPERLION has therefore followed a comprehensive Milestone Program aimed at establishing a platform in Europe for the development of future 3D-MB nano-architecture and nano-fabrication development. The project has therefore providing the enabling Science for a new generation of 3D micro-batteries. In this sense, it is certain to have left its mark on future generations of 3D-MB research activities.

Main dissemination activities
The inevitability of a future international introduction of 3D-MB power sources into the micro-electronics and medical MEMS markets has placed certain social demand on the shoulders of the SUPERLION partners. It has been our underlying task to inform society at different levels of the advances we are instigating in embarking upon the project. Similar Swedish Consortia have a tradition of "never say no" in connection with all contact opportunities with students, engineer clubs, Rotary clubs, school classes, teachers, etc. who approach them with requests to visit our laboratories or listen to us lecture on our visions for the future. Popular-level documentation describing our driving vision and the experimental work within the Consortium has therefore been prepared for schools (students and teachers) and the general public.

For more specific information see the extensive LIST provided of our activities.

Contacts with research society – dissemination of scientific results.
Basic scientific results from SUPERLION have been published through normal procedures for basic research in major international peer-reviewed journals (>30 papers have appeared so far); see the list provided. Much effort has also been to publicize our work at appropriate international conferences and workshops; again, see the list of >60 such dissemination activities.

Creating competence for the future
Many young investigators in the form of PhD students and postdocs have working within SUPERLION (see numbers elsewhere). These have benefitted enormously from the wide range of expertise amongst the more senior scientists active as research leaders and supervisors in the project. Our vision has benn to educate and motivate, over the period of the project. A new generation of 3D-MD researchers has slowly emerged, who are already making an impression in an active way in the further implementation of our Science and Technology.

List of Websites:
SUPERLION website (public access part): www.superlion.eu
Contact person: Professor Josh Thomas (Coordinator) tel: +46705930369 e-mail: josh.thomas@kemi.uu.se

Related information

Contact

Josh Thomas, (Professor)
Tel.: +46-18-4713763
Fax: +46-18-513548
E-mail
Record Number: 196520 / Last updated on: 2017-03-27
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top