Porous silicon thin films as anodes for lithium ion batteries deposited by co-evaporation of silicon and zinc

Porous silicon thin films as anodes for lithium ion batteries deposited by co-evaporation of silicon and zinc

Accepted Manuscript Porous silicon thin films as anodes for lithium ion batteries deposited by co-evaporation of silicon and zinc Stefan Saager, Bert...

2MB Sizes 0 Downloads 53 Views

Accepted Manuscript Porous silicon thin films as anodes for lithium ion batteries deposited by co-evaporation of silicon and zinc

Stefan Saager, Bert Scheffel, Olaf Zywitzki, Thomas Modes, Markus Piwko, Susanne Doerfler, Holger Althues, Christoph Metzner PII: DOI: Reference:

S0257-8972(18)31274-X https://doi.org/10.1016/j.surfcoat.2018.11.064 SCT 24030

To appear in:

Surface & Coatings Technology

Received date: Revised date: Accepted date:

1 June 2018 12 November 2018 21 November 2018

Please cite this article as: Stefan Saager, Bert Scheffel, Olaf Zywitzki, Thomas Modes, Markus Piwko, Susanne Doerfler, Holger Althues, Christoph Metzner , Porous silicon thin films as anodes for lithium ion batteries deposited by co-evaporation of silicon and zinc. Sct (2018), https://doi.org/10.1016/j.surfcoat.2018.11.064

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Porous Silicon Thin Films as Anodes for Lithium Ion Batteries Deposited by CoEvaporation of Silicon and Zinc

Stefan Saager1*, Bert Scheffel1, Olaf Zywitzki1, Thomas Modes1, Markus Piwko2, Susanne Doerfler2, Holger Althues2, Christoph Metzner1 Fraunhofer Institute for Organic Electronics, Electron Beam and Plasma Technology (FEP), Winterbergstrasse 28, 01277 Dresden, Saxony, Germany Fraunhofer Institute for Material and Beam Technology (IWS), Winterbergstrasse 28, 01277 Dresden, Saxony, Germany

RI

2

PT

1

NU

SC

*correspondence to: Stefan Saager, email: [email protected] phone: +49 351 2586-316 fax: +49 351 2586-55316

MA

Abstract

Porous silicon thin films were fabricated by an innovative method using vacuum processing

D

for the application as high capacity anode in lithium ion batteries. The deposition procedure

PT E

comprises a co-evaporation of silicon and zinc, resulting in a deposition of a compound layer with deposition rates up to 100 nm/s and a subsequent thermal annealing. Due to its high

CE

vapor pressure, the zinc fraction is expelled and hence, a porous silicon matrix is formed. Herein, we introduce a novel and potentially scalable synthesis method for porous silicon

AC

films and show first analytical investigations concerning the layer morphology and the electrochemical properties. With the novel silicon anode excellent electrochemical performance, particularly high capacities of ≥3000 mAh/g, reasonable coulombic efficiencies of ≥90 % in the initial cycle and comparably high cycle life >150 cycles can be demonstrated, which reveals their great potential for battery anode applications.

1

ACCEPTED MANUSCRIPT

SC

RI

PT

Graphical abstract

Highlights

Deposition of silicon-zinc compound layers with deposition rates up to 100 nm/s



Subsequent thermal annealing in vacuum for expelling zinc and generating pores at the locations of former zinc grains



Assembling in CR2016 in a half cell configuration using a lithium counter electrode



Encouraging results in electrochemical performance show high capacities ≥3500 mAh/g, initial coulombic efficiencies ≥90 %, and comparably high cycle life >150 cycles

CE

PT E

D

MA

NU



AC

Keywords: porous silicon, silicon zinc compound, electron beam evaporation, lithium ion battery

2

ACCEPTED MANUSCRIPT 1. Introduction Currently, there is an ongoing need for small, efficient and cheap autarkic power supplies with sufficient cycle stability for electric vehicles as well as for transportable electronics (e.g. smartphones, laptops and wearable devices). Lithium-ion batteries (LIB) are well established, but they still offer a high potential for improvements concerning high energy density, low

PT

self-discharge properties and long lifetime. In commercial cells, specific energy densities up to 250 Wh/kg or 700 Wh/l, respectively, are already feasible. A further increase of the energy

RI

density is limited mainly due to material properties of the used electrodes [1]. In commercial

SC

LIB replacing graphite by silicon (Si) as anode material could enhance performance given that Si shows a nearly ten times higher theoretical capacity of 3579 mAh g-1 in the fully lithiated

NU

state at room temperature (Li15Si4) compared to 372 mAh g-1 for graphite. Besides, comparing

MA

to lithium, the element Si is an abundant material in the earth's crust, it is non-toxic, chemical stable, and therefore, it can be mined industrially in large quantities with low cost. However, a compact Si anode is not practicable for LIB. During the lithiation of silicon an intermetallic

PT E

D

phase (LixSi) emerges, which results in a volume expansion over 300 % in the fully lithiated state accompanied with amorphization of the anode material. During periodic charging and discharging cycles, the extensive breathing of Si generates critical stress, which leads to

CE

pulverization, loss of the electrical contact from the current collector, both resulting in a fast

AC

capacity decay of the anode. Furthermore, the solid electrolyte interphase (SEI) becomes instable during continuous cycling consuming irreversible the lithium active material of the cathode at the anode leading to cell-dry- out [2]. In order to overcome the drawbacks, several concepts for Si anode structuring were pursued in past. They include the deposition of amorphous silicon layers with reduced film thickness [3], using structured substrates [4], deposition of Si nanoparticles [5] implemented in binders [1], the growth of Si nanowires [6] and the generation of porous Si by galvanic treatment [7,8]. Mentioned methods show

3

ACCEPTED MANUSCRIPT promising results but they suffer from complex and expensive production process or use toxic substances.

2. Approach

PT

Contrary to mentioned methods, we use a bottom-up approach to build up a porous Si matrix based on vapor phase dealloying. Dealloying in general is a common process during which an

RI

alloy is ‘parted’ in separate components and one element is removed selectively. In a

SC

conventional dealloying process, the initial matrix consists of an alloy with homogenous distribution of the elements, in which one or more of the components is removed by chemical

NU

or electrochemical etching [9], by dissolving in a metal melt [10] or by subliming in vacuum

MA

[11, 12], respectively. To get a porous matrix from it, the remaining component has to reorder its structure by surface diffusion mechanism [13]. The pore size and its spatial distribution

D

depends drastically on the reorder mechanism and can be controlled only in a very limited

PT E

range. Nevertheless, nano-porous Si powder, which was prepared by using a Mg-Si alloy powder, shows already encouraging battery performance. A porous Si matrix was fabricated by immersing the alloy into a Bi melt and dissolve Mg in it [10] or by expelling Mg by

CE

vacuum distillation via thermal annealing process [12], respectively. For further improvement

AC

by tuning porosity and pore size to the demands of applications new fabrication concepts are required, which have the potential for cost-effective mass production with ecological footprint. Recently, Han et al. published a vapor phase dealloying approach, which uses a zinc based binary alloy precursor and takes the advantage of significant difference in vapor pressure [14]. Two types of precursor material were fabricated to small ribbons by melt spinning: a single phase Ni-Zn alloy with homogenous element distribution and a bicontinuous Ge-Zn alloy, congregating separate phases of pure Ge and Zn. Investigations of a bicontinuous Ge-Zn alloy show some evidences for manipulating the porosity caused by 4

ACCEPTED MANUSCRIPT varying the grain size and the content of the immiscible elements. To fabricate low cost efficient battery anodes this method suffers from the inappropriate material choice with Ge and it suffers from their need for further processing steps like milling and sintering the porous material to a flat geometry. Contrary, we applied a sophisticated approach forming a thin film of a bicontinuous alloy

PT

precursor by co-deposition of zinc (Zn) and silicon. The mutual solubilities of Zn and Si in the solid state are negligible [15] and separated grains of pure elements are formed during

RI

layer growth (Fig. 1 (a)). As shown in Fig. 2, the vapor pressure curves and melting points of

SC

Zn [16] and Si [17] differ to several orders of magnitude. Consequently, dealloying by annealing of deposited layers in vacuum at adequate temperatures also leads to a well-known

NU

re-evaporation and expelling of zinc. Pores are generated at the locations of former zinc grains

MA

(Fig. 1 (b)).

The grain size and its distribution depend on deposition process parameters and thus, porosity

D

parameters can be adjusted by adapting deposition parameters. This paper introduces the

PT E

synthesis of porous Si films by the deposition and annealing of Si-Zn compound layers and evaluates their suitability for battery applications. The authors are not aware that the

previously.

CE

deposition of Si-Zn compound layers and subsequent dealloying was investigated in detail

AC

3. Material and methods 3.1 Coating preparation We used 1 mm thick copper sheets (99.9 % purity) as substrates for the physical vapor deposition. For composition and structure analysis sheets of a size of 15x96 mm² and for testing the electrochemical performance copper disks of Ø=14.5 mm were used. Surface was cleaned ex-situ prior deposition in ethanol ultrasonic bath. To prevent alloying by diffusion of zinc into copper, the substrates were pre-coated with a sublayer of stainless steel or chrome, respectively, in a separate vacuum coating tool. Prior 5

ACCEPTED MANUSCRIPT sublayer deposition substrates were cleaned in-situ by a plasma etching process. Therefore, substrates were kept under argon atmosphere of 5 Pa and to ignite plasma glow discharge they were set to a bias of – 900 V for a period of 40 s, which reveals a median copper removal of 13 nm. Afterward substrates were set to floating potential and the deposition of 0.1-1.2 µm interlayer was realized by DC magnetron sputtering using the device PPS 8R with stainless

PT

steel or chrome target, respectively, at a static coating rate of (6.5±0.5) nm/s and at a linear plasma power density on race track of 72 W/cm. The argon pressure was reduced to 8·10-2 Pa

RI

to minimize scattering interaction of vapor particles and to observe a dense and compact

SC

interlayer morphology. Subsequently, the substrates were transferred with vacuum interruption into another vacuum coater equipped with an axial electron beam gun as

NU

schematically shown in Fig. 3.

MA

In order to avoid the propagation of zinc material in the entire recipient the experimental tool was equipped with water cooled shields at which zinc vapor was condensed and collected.

D

A set of 8 sheets or 40 disks, alternatively, were mounted side by side in a movable linear

PT E

transport system.The measurement of substrate temperature was realized by placing stainless steel sheets with rear side welded thermocouples (K-type) between copper substrates. Due to vacuum interruption, the samples were cleaned in-situ by moving through the plasma region

CE

of a magnetron sputter etching tool [18] (left process station in Fig. 3), resulting in a median

AC

surface removal of (40±10) µg/cm2. After pretreatment, substrates were heated up by radiation heater at the right process station depicted in Fig. 3. Samples with an initial temperature in the range of 20-200 °C were moved to the coating position for a defined period. Static deposition process was realized by the electron beam evaporation of zinc and silicon from graphite crucibles with a fast deflected electron beam in high vacuum at 3·10-3 Pa. The beam power (412 kW) and the duty cycle of beam deflection determine the evaporation rates and hence, the composition of intermixed vapor stream. Parallel depositing of multiple samples should offer the possibility to variate samples properties under ensuring comparable process parameters. 6

ACCEPTED MANUSCRIPT The actual evaporation rate of each crucible was determined by two in-situ weight measurement systems [19] and was controlled separately by adapting the e-beam parameters. 4-8 µm thick Si-Zn layers with varying composition were deposited. After deposition process, the substrates were transferred back to the heating station and were annealed to temperatures between 100-550 °C for a time of 10-120 min for zinc re-

PT

evaporation. Samples’ weight was measured with Sartorius Laboratory LC 1200s with an accuracy of 0.01 mg prior and after each process step followed by an evaluation of plasma

RI

etching removal and layer mass loadings, respectively. Additionally, layer thickness was

SC

determined by Dektak profilometer at a layer step obtained on a partially shadowed substrate.

NU

3.2 Characterization of layer morphology and composition

MA

Cross section preparation was carried out by ion preparation technique with JEOL SM-09010. For scanning electron microscopy, a Hitachi SU8000 with a field emission gun was used. The

D

material contrast of backscattered electrons (BSE) was analyzed to distinguish between

PT E

silicon and zinc phases, and the surface topography was imaged using secondary electron (SE) signal. The chemical composition was analyzed by energy-dispersive X-ray

CE

spectroscopy (EDS) with an EDAX Apollo XV system.

AC

3.3 Electrochemical characterization Electrochemical tests were carried out in a half cell configuration in CR2016 coin cells (MTI Corp.). Within an argon atmosphere glovebox (MBraun glovebox, <0.1 ppm O2 and H2O) the coated samples were assembled using a lithium counter electrode (MTI Corp., 99.0 %, diameter: 15.6 mm, thickness: 250 mm). Both electrodes were separated by a polyethylene separator. 30 µl of LP30, a mixture of 1M LiPF6 in EC:DMC (1:1, v:v), was used with 5 v-% FEC additive as electrolyte purchased by Solvionic. For the cells with capacity limitation, a glass fiber separator (Whatmann), and 80 µl of LP30 electrolyte were used. All cells were 7

ACCEPTED MANUSCRIPT electrochemically tested with a BaSyTec Cell Test System (CTS) at (24±2) °C. For the tests with full lithiation and delithiation, the cells were galvanostatically charged and discharged with a current density of C/10 after one formation cycle with C/40 in the cell voltage range between 1 V and 10 mV. In another test, the lithiation capacity of silicon was limited to 2000 mAh/gSi by maximum time for lithiation. The current rates and the maximum potential

PT

range (maximum lithiation to 10 mV) was not changed. The current rate always was related to theoretical anode capacity, calculated from the determined mass loading and theoretic specific

SC

RI

capacity of silicon.

4. Results

NU

4.1 Composition

MA

Substrate area located directly over crucible were coated with the deposition rate 𝑑̇0 . However, the vapor flux density Φ(𝜗) of each vapor source depends on the angle 𝜗 between

D

the vapor propagation direction and the normal of the vapor-emitting surface and is

PT E

proportional to a cos 𝕟 𝜗 term. By using geometrical relationship between vapor emitting surface and vapor collecting surface the growth rate 𝑑̇ of film thickness on a plane substrate is

detail at [20].

CE

proportional to the cos3 𝜗 term in Eq. 1 and is proportional to Φ(𝜗), which is described in

Eq. 1

AC

𝑑̇(𝜗) ∝⋅ cos3 𝜗 ⋅ Φ(𝜗)

Consequently, the deposition rate 𝑑̇ (𝑥S ) from one vapor source differs for lateral position 𝑥S (𝜗) of side by side arrangement of substrates according to the generally valid Eq. 2. 𝑑̇(𝑥S ) = 𝑑̇0 ⋅ cos𝕟+3 𝜗(𝑥S ) Eq. 2 The symbols in Fig. 4 represent the measured static deposition rates for silicon (blue triangles), zinc (red squares) and silicon-zinc compound (purple diamonds), respectively, versus side by side sample position. For better comparability a normalized scale was chosen. A static deposition rate of 100 nm/s was obtained at the center of substrate arrangement in 8

ACCEPTED MANUSCRIPT case of Si-Zn co-deposition. An approximation of data could be achieved by curve fitting according to Eq. 2. Best fits were found for evaporation of silicon by 𝕟Si = 6, for that of zinc by 𝕟Zn = 2. Consequently, the rate distribution results in a variation of the concentration for Si and Zn. Fig. 5 represents concentration data for Si and Zn measured by EDS.

PT

4.2 Morphology As expected the substrate temperature and the zinc concentration proved to be the major

RI

factors influencing the layer morphology. Due to process heat (caused by latent heat of

SC

condensation, thermal radiation from crucibles and backscattered electrons) the substrate temperature rises from TStart almost linear during deposition, as shown in Fig. 6. Enhancing

NU

initial substrate temperature by preheating leads to an equivalent shifted curve (red dashed

MA

line in Fig. 6).

The influence of a varying Zn concentration on the topography of coatings as deposited is

D

shown in Fig. 7 (a,b). Samples were prepared at the same deposition experiment with constant

PT E

evaporation rates for Si and Zn but at different substrate position. The corresponding cross sections are shown in Fig. 8. The distribution of Zn and Si could be visualized by material contrast. Increasing Zn content (compare Fig. 8a versus 8b) results in a prevalent generation

CE

of bumps and a roughening of the layer surface. During layer growth, the substrate

AC

temperature rises according to black solid curve in Fig. 6. After reaching critical parameter concerning Zn concentration and substrate temperature, Zn grains begin to agglomerate with grain sizes up to several µm. This can be explained by their tendency to minimize their internal energy. These Zn grains are the origin for building up bump defects. Fig. 7 (c,d) and Fig. 8 (c,d) show samples from another deposition experiment at equivalent substrate position with increased initial substrate temperature according to red dashed curve in Fig. 6. The increased initial temperature results in a smoother surface, in smaller Zn grains (<100 nm) with a more fine dispersed distribution and in a reduced Zn concentration. 9

ACCEPTED MANUSCRIPT

4.3 Annealing The feasibility for removing Zn was evaluated by vacuum annealing of the Si-Zn compound layer immediately after deposition without vacuum interruption. Fig. 9 illustrates determined

PT

mass loadings 𝑚/𝐴 (i.e., ratio of layer mass 𝑚 and coated area 𝐴) of pure Si layers deposited without Zn (blue triangles), of silicon-zinc compound layers as deposited (purple diamonds)

RI

and of Si-Zn layers annealed in vacuum (green cycles) versus lateral sample position. The

SC

results shown in Fig. 9 demonstrate that Zn is expelled almost completely and mass loadings of nearly pure silicon layers are obtained after annealing. The corresponding mass density 𝜌

NU

can be calculated from mass loading and the film thickness 𝑑porous for Si-Zn layers or the

MA

film thickness 𝑑bulk for pure silicon bulk layers, respectively, measured by profilometer. According to Eq. 3 the porosity ε was calculated from the ratio of mass densities of dealloyed

𝜌porous 𝜌bulk

=1−

PT E

ε=1−

D

Si-Zn-coatings and bulk Si coatings.

𝑚porous 𝑚bulk

𝑉bulk

⋅𝑉

porous

=1−

𝑚porous 𝑚bulk

𝐴 ⋅𝑑bulk

⋅ 𝐴 ⋅𝑑

porous

Eq. 3

In Fig. 9 the determined porosity values are illustrated by filled red square symbols in

CE

dependency of the lateral sample position. The porosity varies between 20  % and 54  % depending on initial Zn content.

AC

Results of an EDS analysis quantified by red numbers in Fig. 10 with Zn concentrations of 22.4 at-% and 0.2 at-%, respectively, confirm that the Zn was almost expelled completely by annealing. Zn grains, visualized by bright areas in Fig. 10 (a), are absent after vacuum annealing and pores, represented by dark regions in Fig. 10 (b), are generated instead of the former Zn grain locations.

10

ACCEPTED MANUSCRIPT 4.4 Diffusion barrier Coating and annealing experiments on pure copper substrate lead to layer delamination and changed re-exposed substrate interface to a brass shining appearance. For higher substrate temperature the responsible inter-diffusion effect is already detectable after Si-Zn coating (Fig. 11). Zn diffusion into the substrate is unacceptable, because it reorders the interface and

weight and reduces gravimetric capacity of the final device.

PT

so it reduces detrimental layer adhesion. Furthermore, the heavy Zn increases substrate’s

RI

Depositing a sublayer by DC magnetron sputtering, alloying can be avoided and adhesion

SC

properties were significantly enhanced. FeCrNi as well as pure chromium layers were tested as barrier material. Best adhesion results were observed by using of a 1 µm thick FeCrNi

NU

sublayer. But as shown in Fig. 12, diffusion could not prevented completely. Copper material

MA

is transferred through grain boundaries of the barrier layer, whereby voids are generated at the interface and pillars are build up on silicon side. This effect is known as Kirkendall effect

PT E

D

[21].

4.5 Electrochemical properties

CE

The electrochemical properties of prepared porous silicon layers were evaluated by

AC

galvanostatic cycling with lithiation and de-lithiation processes using coin cells in half-cell configuration. In Fig. 13 the results of capacity measurements with full lithiation and delithiation of a fully processed porous silicon film and a compact silicon layer deposited without zinc are shown over 30 cycles. At beginning of cycling, both samples show a high initial capacity over 3500 mAh/g reaching the theoretical value of Li15Si4 phase. The initial coulombic efficiency of the porous silicon film anode (89.0 %) is only lowered by three percent compared to the compact silicon film (92.0 %). During cycling of the compact silicon anode, a fast capacity fade was observed within ten cycles. This behavior is probably caused 11

ACCEPTED MANUSCRIPT by cracking, pulverization and delamination of Si coating during cycling. Contrariwise, in case of the porous Si coating, the capacity fade is much lower and a capacity of 1300 mAh/g could be maintained after 30 cycles. These results demonstrate the beneficial influence of the porous structure on the cycle stability of silicon film anodes. However, further adjustments of the pore size, spatial distribution of pores, and ratio between pores and silicon active material

PT

are needed to further enhance the cycle life at full lithiation and delithiation capacity. The pores in the silicon film of the anodes being discussed above are probably insufficient to

RI

accommodate the full volume expansion of Si during full lithiation. Nevertheless, a limitation

SC

of the volume expansion of Si is needed in order to hamper cell fading. Therefore, half cells were tested by a cycling procedure with limited lithiation of Si to 2000 mAh/gSi. Two Si films

NU

with different loadings and porosity were analyzed with this test procedure, which is in

MA

accordance to the typical cycling behavior of Si anodes in LIB (Fig. 14). For the anode with a medium mass loading of 0.62 mgSi/cm2 the limited lithiation capacity of 2000 mAh/gSi could be reached up to 35 cycles before fading occurs. For the Si anode with the lower mass loading

PT E

D

of 0.46 mgSi/cm2 and higher pore volume, a life time of more than 150 cycles with constant capacity was demonstrated. Compared to the fully lithiation and delithiation, the value of the initial coulombic efficiency is slightly lower (81%). During the further cycles an almost

CE

constant coulombic efficiency over 98 % was reached. These results suggest that a tailored

AC

pore volume related to the lithiation capacity (i.e., volume expansion) of Si is needed to achieve a high cycle life.

5. Discussion 5.1 Temperature regime for composition and morphology Due to reduce battery weight and volume besides to ensure good electrochemical properties, for the final battery application a thin copper foil is predestinated for anode electrode. Feasibility of inline silicon deposition on thin copper foils for battery applications was already demonstrated [22]. Nevertheless, the heat capacity of thin copper foils would be too low for 12

ACCEPTED MANUSCRIPT vapor deposition of some micrometer thick layers with high deposition rate. An industrial roll to roll coating on thin metal foils requires the implementation of an additional substrate cooling equipment with sufficient high cooling efficiency. The availability of such technical solutions was already shown by Heinß et al. [23]. In this study, 1 mm thick copper substrates with higher heat capacity were used in order to avoid too high temperatures during vapor

PT

deposition. In this way, the basic principles of generating porous silicon layers could be investigated with moderate effort. Si-Zn compound layers have not yet been described

RI

elsewhere and therefore, the relations between process parameters and layer morphologies as

SC

shown in Fig. 8 cannot easily be understood without further investigations. The measured distribution of layer thickness (Fig. 4) indicates an unexpected high value of

NU

exponent 𝕟𝑆𝑖 . It is known that 𝕟 may depend on the power density of the electron beam

MA

impinging at the melt [20] and related to high evaporation rates. Since such high rates were not obtained within this study it is assumed that focusing effects caused by the crucible played

D

a dominant role. The experiments were carried out with low level of Si melting pool in the

PT E

crucible. Therefore, the obstruction of vapor propagation by the crucible has also to be taken in to account [20]. This would explain the high value of 𝕟 determined in our experiments. Within Fig. 7 and 8 different Zn concentrations are called for comparable deposition

CE

experiments. It can be assumed, that a remarkable re-evaporation of Zn occurs at temperatures

AC

above 200 °C. The re-evaporation leads to a depletion of Zn already during the deposition, what is consistent with EDS analyses. Circumstances leading to finely dispersed distribution of Zn crystallites and such leading to Zn agglomeration will be of special interest and further experiments with improved experimental setup are planned. Moreover, experiments applying a gradual Zn concentration during layer growth will be designed in order to adjust the porosity and measure it by techniques that are more accurate. Assuming a film volume of 𝑉porous = 3 × 𝑉bulk (i.e., a volume expansion of 300 % in the fully lithiated state Li15Si4) and inserting 𝑉porous in Eq. 3, a porosity of ε  ≈ 66 %  would be required for compensation by withholding 13

ACCEPTED MANUSCRIPT layer thickness expansion. To enhance the cycle stability of commercial battery applications by reducing volume expansion, a partial lithiation is suitable only. This also concurs with the results presented in Fig. 13 and 14. Therefore, a reduced volume expansion and chances of layer thickness has to be taken into account. Consequently, the optimal porosity depends on concrete design of a battery cell, in particular the balancing between anode and cathode

PT

properties. Further electrochemical investigations are necessary for an estimation of an optimal porosity value, which is needed to obtain best cyclability performance of a battery

RI

cell.

SC

Applying the described coating and dealloying procedures in a production line, huge quantities of Zn will be consumed during processing. From productivity and ecology aspects,

NU

an adequate recycling method especially for Zn has to be contrived. The installation of cooled

MA

shieldings at process stations for deposition and for thermal annealing are conceivable. During coating process, some loss by unintentional parasitic deposition on the walls inside the

D

chamber has to be taken into account. The condensed material will consist of silicon and zinc

PT E

and has to be recycled. In contrast, re-evaporated material from annealing station will condense on cooled surfaces as pure Zn only, which can be detached during periodic maintenance and can be reused directly. Therefore, for expelled Zn a recyclability of nearly

AC

CE

99 % is feasible.

5.2 Electrochemical performance The results of the electrochemical tests are very encouraging though the mass loading in the order of 0.6 mgSi/cm2 as used in this study in combination with the used electrolyte is still too low. Estimations show that a Si mass loading of 1.0 mgSi/cm2 is necessary for their application in high energy density lithium ion batteries. Porous Si anodes show an excellent initial capacity of over 3500 mAh g-1, which is very close to the theoretical value of 3579 mAh g-1. In particular, the high coulombic efficiency at first 14

ACCEPTED MANUSCRIPT charging cycle of almost 90 % at first charging cycle is evident for low lithium losses during SEI formation. This is important to sustain high energy density of final LIB application, because consumed lithium is trapped in the SEI, and depleted lithium has to be provided by larger and consequently heavier reserves sustained from cathode material. Electrochemical characterization procedure has a major influence for the evaluation of Si

PT

layers as anode material. Using the entire capacity by charging completely as shown in Fig. 13 leads to a maximum volume expansion and thus, it increases the risk of pulverization.

RI

Therefore, limiting lithiation and reducing capacity as shown in Fig. 14 is a common

NU

SC

procedure in LIB application to enhance cycling durability.

MA

6. Conclusions

An experimental setup was build up for co-evaporation of Si and zinc to deposit a compound layer with static rates up to 100 nm/s. The layer composition was controlled by adapting

PT E

D

electron beam parameters and by monitoring evaporation rates of each crucible. Zn concentration in the range of 10 - 50 at.-% could be systematically adjusted by a lateral arrangement of substrates. Substrate temperature was varied during deposition process and

CE

during the following annealing phase. Depending on process parameters, layer morphology

AC

ranges from a finely dispersed distribution of Zn crystallites with smooth layer surface up to Zn agglomeration with bumps, defects, and a rough surface. By subsequent thermal annealing under vacuum, Zn component could be expelled almost completely and voids with magnitudes diameters from nanometer to micrometer were generated. This demonstrates the feasibility to produce porous Si layers by an innovative approach with vacuum processing. Evaluation of porous Si layers by electrochemical characterization in half cell configuration shows a high capacity in the range of the theoretical value (3579 mAh/gSi) and an improved capacity retention compared to a compact Si film during full lithiation and delithiation. 15

ACCEPTED MANUSCRIPT However, the cycle stability is insufficient for a practical application probably due to too less pore volume in the Si film to accommodate the volume expansion of Si during lithiation. The limitation of the lithiation capacity to 2000 mAh/gSi seems to meet the available pore volume and in result more than 150 cycles without fade could be reached. Concluding, the interim results of ongoing research project are greatly encouraging but offer also space for

PT

optimization concerning thermal management, adhesion, detailed understanding of

AC

CE

PT E

D

MA

NU

SC

RI

morphology formation, and studying their influences to battery performance.

16

ACCEPTED MANUSCRIPT

Acknowledgments The project was funded by the European Union and the Free State of Saxony (funding reference 100275833).

AC

CE

PT E

D

MA

NU

SC

RI

PT

Comment: Please print the logo EFRE nearby acknowledgment!

17

ACCEPTED MANUSCRIPT List of references

[1] Jang Wook Choi and Doron Aurbach, Promise and reality of post-lithium-ion batteries with high energy densities, Nature Reviews Materials 1 (2016): 16013. https://doi.org/10.1038/natrevmats.2016.13

PT

[2] Hui Wu, Gerentt Chan, Jang Wook Choi, Ill Ryu, Yan Yao, Matthew T. McDowell, Seok Woo Lee, Ariel Jackson, Yuan Yang and Liangbing Hu, Stable cycling of double-walled silicon nanotube battery anodes through solid-electrolyte interphase control, Nature nanotechnology, 7 (2012): 310. https://doi.org/10.1038/nnano.2012.35

RI

[3] Juchuan Li, Alan K. Dozierb, Yunchao Lia, Fuqian Yanga and Yang-Tse Chenga, Crack Pattern Formation in Thin Film Lithium-Ion Battery Electrodes, Journal of The Electrochemical Society, 158 (2011): A689-A694. https://doi.org/10.1149/1.3574027

NU

SC

[4] Leimeng Sun, Xinghui Wang, Rahmat Agung Susantyokoa and Qing Zhang, Coppersilicon core-shell nanotube arrays for free-standing lithium ion battery anodes, J. Mater. Chem. A, 2 (2014): 15294-15297. https://doi.org/10.1039/c4ta03188a

MA

[5] Qin Si, K. Hanai, N. Imanishi, M. Kubo, A. Hirano, Y. Takeda and O. Yamamoto, Highly reversible carbon-nano-silicon composite anodes for lithium rechargeable batteries, Journal of Power Sources , 189 (2009): 761-765. https://doi.org/10.1016/j.jpowsour.2008.08.007

D

[6] Candace K. Chan, Hailin Peng, Gao Liu, Kevin McIlwrath, Xiao Feng Zhang and Robert A. Huggins, High-performance lithium battery anodes using silicon nanowires, Nature nanotechnology, 3 (2008): 31-35. https://doi.org/10.1038/nnano.2007.411

PT E

[7] Young-Lae Kim, Yang-Kook Sun and Sung-Man Lee, Enhanced electrochemical performance of silicon-based anode material by using current collector with modified surface morphology, Electrochimica Acta, 53 (2008): 4500-4504. https://doi.org/10.1016/j.electacta.2008.01.050

AC

CE

[8] Enrique Quiroga-González, Jürgen Carstensen and Helmut Föll, Good cycling performance of high-density arrays of Si microwires as anodes for Li ion batteries, Electrochimica Acta, 101 (2013): 93-98. https://doi.org/10.1016/j.electacta.2012.10.154 [9] Jonah Erlebacher, Michael J. Aziz, Alain Karma, Nikolay Dimitrov and Karl Sieradzki, Evolution of nanoporosity in dealloying, Nature, 410 (2001): 450. https://doi.org/10.1038/35068529 [10] Takeshi Wada, Junpei Yamada and Hidemi Kato, Preparation of three-dimensional nanoporous Si using dealloying by metallic melt and application as a lithium-ion rechargeable battery negative electrode, Journal of Power Sources, 306 (2016): 8-16. https://doi.org/10.1016/j.jpowsour.2015.11.079 [11] Yuxia Sun, Yibin Ren and Ke Yang, New preparation method of micron porous copper through physical vacuum dealloying of Cu-Zn alloys, Materials Letters, 165 82016): 1-4. https://doi.org/10.1016/j.matlet.2015.11.102

18

ACCEPTED MANUSCRIPT [12] Yongling An, Huifang Fei, Guifang Zeng, Lijie Ci, Shenglin Xiong, Jinkui Feng and Yitai Qian, Green, Scalable, and Controllable Fabrication of Nanoporous Silicon from Commercial Alloy Precursors for High-Energy Lithium-Ion Batteries, ACS Nano,12 (2018) 4993-5002, 2018. https://doi.org/10.1021/acsnano.8b02219 [13] Ian McCue, Ellen Benn, Bernard Gaskey and Jonah Erlebacher, Dealloying and Dealloyed Materials, Annual Review of Materials Research, 46 (2016): 263-286. https://doi.org/10.1146/annurev-matsci-070115-031739

PT

[14] Jiuhui Han, Cheng Li, Zhen Lu, Hao Wang, Zhili Wang, Kentaro Watanabe and Mingwei Chen, Vapor phase dealloying: A versatile approach for fabricating 3D porous materials, Acta Materialia, 163 (2019): 161-172. https://doi.org/10.1016/j.actamat.2018.10.012

RI

[15] R.W. Olesinski and G. J. Abbaschian, The Si-Zn (Silicon-Zinc) system, Bulletin of Alloy Phase Diagrams, 6 (1985): 545-548.

NU

SC

[16] C. Alcock, V. Itkin and M. Horrigan, Vapour Pressure Equations for the Metallic Elements: 298-2500K, Canadian Metallurgical Quarterly, 23 (1984): 309-313. https://doi.org/10.1179/000844384795483058

MA

[17] O. Kubaschewski and C. Alcock, Metallurgical thermochemistry, fifth ed., Pergamon press Oxford, 1979.

D

[18] Jörg Faber, Untersuchungen zum Sputterätzen metallischer Substrate mit gepulsten und ungepulsten Magnetronentladungen, Otto-von-Guericke-Universität Magdeburg, dissertation (2000).

PT E

[19] Bert Scheffel and Klaus Goedicke, In situ-force measurement for the determination of the evaporation rate with high-rate electron beam evaporation, Surface and Coatings Technology, 98 (1998): 944-947. https://doi.org/10.1016/S0257-8972(97)00314-9

CE

[20] Siegfried Schiller, Ullrich Heisig and Siegfried Panzer, Vapor Propagation, in: Electron beam technology, second ed., Verlag Technik, 1995, pp. 173-185 ISBN 3-341-01153-6

AC

[21] Hideo Nakajima, The discovery and acceptance of the Kirkendall Effect: The result of a short research career, JOM, 49 (1997): 15-19. https://doi.org/10.1007%2FBF02914706 [22] Markus Piwko, Thomas Kuntze, Sebastian Winkler, Steffen Straach, Paul Härtel, Holger Althues and Stefan Kaske, Hierarchical columnar silicon anode structures for high energy density lithium sulfur batteries, Journal of Power Sources, 351 (2017): 183-191. https://doi.org/10.1016/j.jpowsour.2017.03.080 [23] Jens-Peter Heinß, Peter Lang and Patrick Ruppelt, Temperature control of metal strip during high-rate vacuum coating, Surface and Coatings Technology, 290 (2016): 39-42. https://doi.org/10.1016/j.surfcoat.2015.08.028

19

ACCEPTED MANUSCRIPT Tables

AC

CE

PT E

D

MA

NU

SC

RI

PT

No tables are included in this manuscript.

20

ACCEPTED MANUSCRIPT List of figure captions Figure 1. Scheme of the formation of separated grains of zinc (red) and silicon (blue) after deposition (a) and porous silicon film after expelling zinc by thermal annealing (b).

Figure 2. Temperature dependencies of the saturated vapor pressure of zinc (red, dashed line)

PT

[16] and silicon (blue, solid line) [17]. The symbols show material’s melting temperature.

RI

Figure 3. Scheme of the experimental setup for deposition of silicon-zinc layers and vacuum

SC

annealing.

NU

Figure 4. Distribution of deposition rates for Si and Zn along the lateral position,

MA

experimental data (symbols) together with fits (lines).

D

Figure 5. Concentrations of silicon and zinc in terms of their distribution along the sample

PT E

position measured by EDS with a relative measurement uncertainty of 5  %.

CE

Figure 6. Substrate temperature during depositing 5-8  µm thick silicon-zinc compound layer.

AC

Figure 7. Topography of Si-Zn compound layers imaged with SE signal of SEM for low (a,b) and medium (c,d) substrate temperature during deposition. The colored numbers represent Si (blue) and Zn (red) concentration determined by EDS.

Figure 8. Cross sections of silicon-zinc compound layers corresponding to Fig. 7, imaged by SEM with material contrast.

21

ACCEPTED MANUSCRIPT Figure 9. Mass loading (open symbols) vs. lateral substrate position for different process steps. The corresponding porosity (filled symbols) for annealed samples was derived from layer thickness and mass loading.

Figure 10. Material contrast cross section of Si-Zn compound layers with details near the

PT

surface before (a) and after (b) annealing in vacuum.

RI

Figure 11. Material contrast cross section with details near substrate interface after depositing

SC

3 µm thick layer of Si-Zn compound. The bright alloy region below interface results from zinc

NU

diffusion into the copper substrate.

MA

Figure 12. Cross section of a FeCrNi diffusion barrier layer after Si-Zn deposition and

D

vacuum annealing.

PT E

Figure 13. Capacity development of compact and porous silicon thin film measured in a halfcell vs. lithium metal, fully lithiated at initial cycle.

CE

Figure 14. Capacity development and coulombic efficiency of porous silicon thin films

AC

measured in a half-cell vs. lithium metal with limited lithiation capacity of silicon.

Figure Acknowledgments (without any captions)

22

ACCEPTED MANUSCRIPT

PT

Figures

MA

NU

SC

RI

Fig. 1

AC

CE

PT E

D

Fig. 2

23

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

Fig. 3

Fig. 4 (changed by adding error bars)

24

PT

ACCEPTED MANUSCRIPT

D

MA

NU

SC

RI

Fig. 5 (changed by adding error bars)

AC

CE

PT E

Fig. 6

Fig. 7 (changed by annotation style)

25

PT

ACCEPTED MANUSCRIPT

MA

NU

SC

RI

Fig. 8 (changed by annotation style)

AC

CE

PT E

D

Fig. 9 (changed by adding error bars and by adding porosity values)

Fig. 10 (changed by annotation style)

26

ACCEPTED MANUSCRIPT

AC

Fig. 13

CE

PT E

D

MA

Fig. 12 (changed by annotation style)

NU

SC

RI

PT

Fig. 11 (changed by annotation style)

Fig. 14 27