Lithiation and Delithiation Mechanisms of Gold Thin Film Model Anodes for Lithium Ion Batteries: Electrochemical Characterization

Electrochimica Acta 164 (2015) 81–89

Contents lists available at ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

Lithiation and Delithiation Mechanisms of Gold Thin Film Model Anodes for Lithium Ion Batteries: Electrochemical Characterization P. Bach a,b,c, M. Stratmann b,c , I. Valencia-Jaime d,e , A.H. Romero e , F.U. Renner a,b, * a

Institute for Materials Research (IMO), Hasselt University,Wetenschapspark 1, 3590 Diepenbeek, Belgium Max-Planck-Institut für Eisenforschung GmbH, Max-Planck-Straße 1, 40237 Düsseldorf, Germany Center for Electrochemical Sciences — CES, Universität Bochum, Universitätsstraße 150, 44780 Bochum, Germany d Centro de Investigacion y Estudios Avanzados del IPN, MX-76001 Santiago de Querétaro, Mexico e Physics Department, West Virginia University, White Hall, Box 6315, WV-26506-6315 Morgantown, USA b c

A R T I C L E I N F O

A B S T R A C T

Article history: Received 3 January 2015 Received in revised form 21 February 2015 Accepted 22 February 2015 Available online 24 February 2015

Lithium Ion batteries have to be significantly improved to fulfill the challenging needs in electromobility or large scale energy storage technology. In this context the use of model electrodes such as single-crystals or thin films allows well-defined mechanistic studies. Here we present a detailed electrochemical investigation of the lithiation–delithiation behavior of Au thin film model electrodes in ionic liquid electrolyte. Cyclic voltammetry, galvanostatic-, stepwise potentiostatic lithiation–delithiation cycles, as well as galvanostatic intermittent titration technique, GITT, measurements were performed. We found nearly identical mechanism of Li insertion and extraction in these three types of measurements at different current levels. The mechanism of alloying or lithiation deviated from the mechanism of dealloying or delithiation. While during the lithiation process two main plateaus related to phase transformations occur in the potential–time curves three main plateaus appear during delithiation. First results of theoretical simulations confirm a high number of possible metastable phases in the Li–Au system. The measurements also point to the influence of SEI-film formation on the cycling behavior. Based on these insights a mechanistic sequence and a phase evolution diagram for the electrochemical alloying of Li with Au are presented. ã 2015 Published by Elsevier Ltd.

Keywords: electrochemistry lithium ion batteries Au model anodes lithiation–delithiation cycles

1. Introduction Interfacial electrochemistry has often used Gold (Au) substrates as model systems for various aspects of surface science problems, from metal-organic interfaces, such as for example thiol selfassembled monolayers [1,2], (bio)-sensor applications [3–6] or fundamental oxidation studies [7]. The high inertness and ease of sample preparation by flame annealing renders Au an appealing model system for electrochemical studies. Also in Li-ion battery (LIB) research Au is used as coating [8] or (nanoporous) template [9], as well as as model system for bulk electrodes [10,11]. Au seed layers are furthermore employed for Si nano-wire growth for LIB Si anodes and Au is thus an integral part of the respective LIB anode [12]. Lithium (Li) as a small and reactive element diffuses readily in Au bulk materials (similar to Zn [13]), which has been recently employed to obtain Au–Li alloy precursors for further

* Corresponding author at: Institute for Materials Research (IMO), Hasselt University, Wetenschapspark 1, 3590 Diepenbeek, Belgium. Tel.: +32 11 26 8835. E-mail address: [email protected] (F.U. Renner). http://dx.doi.org/10.1016/j.electacta.2015.02.184 0013-4686/ ã 2015 Published by Elsevier Ltd.

producing high-surface-area nanoporous gold (npg or npAu) by dealloying [14]. The dealloying approach has been early used to produce Raney Ni [15] or more recently dealloyed core-shell nanoparticles for catalysis [16]. But npAu and other porous elements have in the last years been proposed for a number of further potential applications from actuators and bio-sensors to Li-ion batteries and supercapacitors [9,17]. Such porous electrodes do better cope with the respective immense volume changes involved in alloying reactions. For all these topics a thorough understanding of the chemical, structural and microstructural evolution during alloying and dealloying cycles is necessary. Electrochemistry embodies mainly integral techniques measuring the changes in an external electrical current or potential originating from macroscopic surfaces. Nevertheless, in combination with well-ordered thin films or single crystal surfaces even atomic scale information can be obtained due to the low electrical currents which can be technically measured and the relatively large effects on potential which even small changes of the solid-electrolyte interface structure may have. Examples are the monolayer adsorbate structures of electrolyte molecules such as sulfate on Au(111) [1,18] or sub-monolayer changes during multistep under-potential

82

P. Bach et al. / Electrochimica Acta 164 (2015) 81–89

deposition (UPD) such as Cu on Au (111) [19]. But it is also clear that in most cases additional complementary, potentially in-situ, microscopic information is needed to clarify the true structures and mechanisms [20,21]. Prominent contributions have been hereto provided by electrochemical scanning tunneling microscopy (EC-STM) revealing local atomic and molecular interface structures [21] or by the use of Synchrotron radiation for X-ray absorption spectroscopy (XAS and EXAFS) and in-situ X-ray diffraction [22–25]. This project is part of a more extended ongoing Synchrotronbased in-situ X-ray diffraction study on lithiation and delithiation of pure Au anodes mimicking a model battery charging and discharging process. Next to the inertness of the material, the high cross-section for X-ray scattering renders Au an ideal substrate. The detailed diffraction results will be reported separately. While for frequently time consuming electrochemical diffraction studies often potentiostatic modes are employed, batteries are usually characterized by galvanostatic cycling. Therefore we here thoroughly compare potentiostatic and galvanostatic modes and report next to more common cyclic voltammetry and galvanostatic cycling also less frequently used Galvanostatic Intermittent Titration Technique (GITT) measurements [26].

3. Results and Discussion All electrochemical measurements presented here have been obtained from a sputter-deposited 100 nm thin Au film on a polycrystalline Cu substrate. The electrolyte was a 0.3 M solution of LiTFSI in Pyr14TFSI and the reference electrode commercial Li foil.

2. Experimental The sample for the presented electrochemical measurements was a 100 nm thin Au-film on a polycrystalline Cu substrate. Cu discs (8 mm diameter, 0.5 mm thickness) were first ground (grain sizes 1000 and 2500) and then polished (using silica particles of 50 nm grain size in basic aqueous suspension). The Au film was applied by RF magnetron sputtering in Ar atmosphere at room temperature at a deposition rate of 1 Å/s. The Au target was supplied by Kurt J. Lesker Company Ltd., England, and the deposition was controlled by an Inficon IC6 deposition controller (Inficon, Switzerland). The electrochemical measurements were carried out in a custom made electrochemical Teflon cell allowing a spot of the sample of 4 mm in diameter to be in contact with the electrolyte. Li foils (Kisco Ltd. Japan) served as counter and reference electrode. The electrochemical cell was assembled and operated inside an Ar-filled glovebox (SylaTech GmbH, Germany) with O2 and H2O contents below 1 ppm. The electrolyte used was a 0.3 M solution of the Li-salt LiTFSI (Merck KGaA, Germany) in the ionic liquid Pyr14TFSI (1-butyl-1-methylpyrrolidinium-bis(trifluoromethanesulfonyl)-imide) [27] (Solvoionics, France). We used an Ivium Compact Stat potentiostat (Ivium Technologies B.V., Netherlands). CVs were acquired between 1 V and 0 V vs. Li/Li+ at a potential scan rate of 1 mV/s. Galvanostatic lithiation–delithiation cycles were performed at different currents. At 1.7 C (16.6 mA) the cell was cycled between 1.0 V and 5 mV and at 0.17 C (1.66 mA) between 1.5 V and 5 mV. Galvanostatic Intermittent Titration Technique (GITT) measurements [26] were conducted at 0.17 C alternating between 1000 s of galvanostatic lithiation and delithiation and 2000 s at open circuit potential (OCP) at each step. Potentiostatic lithiation–delithiation cycles in a potential range between 1.5 V and 5 mV were performed by stepwise lowering and raising the potential to avoid large currents. The next potential step was triggered with the current undercutting a set respective limit. In case of the measurements employing a 5 mA current limit the step size was 0.1 V between 1.5 V and 0.3 V during lithiation and between 0.5 V and 1.5 V during delithiation. In the potential range below the potential step size was 20 mV. Employing a current limit of 1 mA the potential step size was 0.1 V between 1.5 V and 0.3 V during lithiation, between 0.5 V and 1.5 V during delithiation, and 10 mV otherwise.

Fig. 1. (a): Cyclic voltammograms of a 100 nm thin Au film in the voltage range from 0 V to 1.5 V vs. Li/Li+. (b): Pairs of peak-potential and peak-current values of the three anodic peaks from Fig. 1a extracted from the first five CV cycles (squares). For the first anodic peak the values are given starting from the third cycle. The circles represent the respective values of the galvanostatic lithiation–delithiation cycles from Fig. 2a. (c): Lithiation-delithiation efficiency during CV measurements. The letters in brackets denote the associated scale: (l): left, (r): right. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

P. Bach et al. / Electrochimica Acta 164 (2015) 81–89

83

3.1. Cyclic Voltammograms Fig. 1a shows cyclic voltammograms (CVs) in the voltage range from 0 V to 1.5 V vs. Li/Li+. In the first cycle starting from 1.5 V in cathodic direction the current increases in the voltage range below 0.1 V (range A in Fig. 1a) reaching
27 mA at the vertex potential of 0 V. In the anodic branch two oxidation peaks occur at 0.356 V with 6.1 mA and at 0.476 V with 10.4 mA (range B) while up to 1.5 V no further anodic peaks are visible (range C). In the second cycle the increase of cathodic current already starts close to 0.2 V (range A). At the vertex potential the current has increased to
47 mA. Similar to the cathodic branch also the peaks in the anodic branch are higher in maximum current in the second cycle and slightly shifted in peak potential. In the third cycle there is hardly any change in the potential at which the cathodic current starts arising (0.2 V as in the second cycle), but the current reached at the vertex potential is even higher (
49.8 mA) than in the second cycle. The anodic branch in the third cycle shows an additional peak at 0.238 V and 2.9 mA. In the following cycles there are again a single cathodic current peak and three anodic current peaks. In addition to the pronounced peaks in the CV there is a small cathodic feature shown in the inset of Fig. 1a at 0.56 V, most pronounced in the first cycle and subsequently vanishing. Until the eighth cycle the main peak currents and – in case of the anodic peaks – also the peak potentials increase. The respective current and potential of these three anodic peaks of the first five CVs (squares) are plotted in Fig. 1b. The circles represent the current-plateau-potential pairs of values from the galvanostatic lithiation–delithiation cycles (compare Fig. 2a). The oxidative peaks in the anodic branch below 0.7 V represent the opposite process to the reductive process; hence a reversible reductionoxidation step occurs near the Li/Li+ potential. Since decomposition of the electrolyte is an irreversible process, the reversible peaks in Fig. 1a represent thus the lithiation and delithiation of the electrode material. From the integrated charge of the total reductive current (I < 0 mA) and the total oxidative current (I > 0 mA) for the different cycles the nominal lithium content and the efficiency of the Li reduction/oxidation process was calculated (in equivalents compared with the molar quantity of Au atoms and assuming all the charge converted is used for Li reduction). The amount of charge converted during the reductive process (“lithiation”) increases by 57% within the first three cycles (Fig. 1c). In the subsequent cycles there is only a slight further increase noticeable. The amount of charge converted during the oxidative process (“delithiation”) is in the first cycle only 27% of the amount of charge converted during the reductive process. During the following cycles the amount of charge converted during the oxidative process increases at much lower rate. In the 8th cycle it reaches 92% of the charge quantity converted during the reductive process and stays nearly constant during the following cycles. The difference between the two curves discussed (“Li excess”; in Li equivalents) accordingly decreases considerably within the first eight cycles. The cycling efficiency (“efficiency”) as the quotient of the two curves accordingly increases pronouncedly within the same cycle range. The nominal lithium content (“degree of lithiation”; in Li equivalents) follows qualitatively the evolution of charge converted during the oxidative process. Compared to typical commercial battery anode materials Au shows thus a poor cycling efficiency. 3.2. Galvanostatic lithiation–delithiation cycles Fig. 2a shows the potential evolution during the first and second galvanostatic lithiation–delithiation cycle. The original time scale was converted into the nominal Li content x (in LixAu) using the

Fig. 2. (a): First two galvanostatic lithiation–delithiation cycles of a 100 nm thin Au film in the voltage range from 0 V to 1.0 V vs. Li/Li+ at a rate of 1.7 C. (b): Comparison of galvanostatic lithiation–delithiation cycles at two different charging rates of 1.7 C (16.6 mA) and 0.17 C (1.66 mA). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

respective galvanostatic current and time. The third and higher lithiation–delithiation cycles (not shown) were very similar to the second cycle. The first cycle coincides with the subsequent ones starting from the beginning of the second lithiation plateau. For clarification the first cycle is here plotted with a small shift to match the second plateau of the higher cycles. At the beginning of the galvanostatic cycles the potential drops very quickly to a minimum (dip, D10 in Fig. 2a) and then increases slightly to form a first lithiation potential plateau (A10 ). The initial potential decrease forms a dip down to 66 mV in the first cycle (D10 ), and to 180 mV in the second cycle (D1). The first plateau regions (A10 and A1) develop at a potential of 160 mV and 190 mV, respectively, and last approximately to a Li content in LixAu of x = 1. For both cycles after a smaller dip (D2) a second plateau at 60 mV (A2) develops. The second plateau lasts until at a nominal Li content of about 2.5 is reached. At a lower current limit of 5 mV the current direction was reversed to force extraction of Li from the electrode. During the delithiation branches three potential plateaus are visible. The first one at a potential of about 200 mV (B1 in Fig. 2a), a shorter plateau at about 400 mV around a nominal Li content of x = 1 (B2), and the third one at a potential of about 500 mV (B3). The second plateau during delithiation is considerably shorter than the first one and also much shorter than the insertion plateaus. The individual cycles were stopped after reaching an upper potential limit of 1.0 V. Fig. 2a shows that the major differences between the first and the second lithiation–delithiation cycles are only present

84

P. Bach et al. / Electrochimica Acta 164 (2015) 81–89

before the second lithiation plateau starts. Also visible is that obviously even in the higher cycles the delithiation half-cycle of the galvanostatic lithiation–delithiation potential curve is shorter than the lithiation half cycle. Apart from the five obvious potential plateaus in the second cycle some smaller features are worth mentioning. Between the first and the second lithiation plateau (IA1), and between the first and the second delithiation plateau (IB1), as well as in smaller extend also during the final potential increase (IB3) there are less pronounced additional intermediate plateaus (see inset in Fig. 2a). In addition, the initial potential drop in the first cycle shows a shoulder (S). Also the galvanostatic measurements point to the prevalence of reversible lithiation–delithiation near the Li/Li+ potential. Plateaus occurring in galvanostatic potential–time curves usually denote two-phase regions where one phase is transformed successively at the assigned transformation potential into the other phase. The potential plateaus observed in Fig. 2a are therefore assigned to occurring phase transformations of Li–Au alloys. The intermediate plateaus may occur due to some quickly proceeding minor changes in the crystal structure of the alloy phases consuming only little charge. In addition to the galvanostatic measurements at 1.7 C also similar measurements at 0.17 C, i.e. at a tenth of the first rate, were conducted. Fig. 2b shows a comparison of the respective, already stable 4th cycles. Comparable to 1.7 C the measurements at 0.17 C show two main potential plateaus during lithiation (A1, A2 in Fig. 2b) and three during delithiation (B1, B2, B3). The intermediate plateaus between the first and the second lithiation plateau (IA1) and the first and the second delithiation plateau (IB1) as well as during the final potential increase (IB3) are more pronounced at the lower rate. Beyond these particular features the two potential–time curves are similar. As expected from ohmic and electron transfer resistances the plateau potentials are shifted in comparison to the measurements at 1.7 C due to a current-dependent overpotential, i.e. during lithiation the plateau potentials are slightly higher at lower rate and vice versa. Mainly during lithiation also diffusion overpotential may come into effect. Fig 2b also shows the maximum Li content during lithiation with almost 3.0 versus 2.5 equivalents to be higher at 0.17 C than at 1.7 C. Also the extracted value during delithiation is higher at 0.17 C (2.56) than at 1.7 C (2.27). In consequence the discrepancy between the inserted and extracted values is higher at 0.17 C. 3.3. Galvanostatic lithiation–delithiation cycles at 0.17 C with intermediate relaxation periods (GITT) In addition to the galvanostatic measurements also galvanostatic potential relaxation measurements (GITT) [26] were conducted at 0.17 C. The resulting curves are reproduced in Fig. 3a and b. As a main difference to the simpler galvanostatic measurements, GITT measurements feature intermediate periods at OCP for potential relaxation. A similarity of the observed potential–timecurve in Fig. 3a to the potential–time-curve from the purely galvanostatic measurements is apparent. Two potential plateaus during lithiation (A1 and A2) and two apparent potential plateaus during delithiation (B1 and B3) are clearly distinguishable at 230 mV and 110 mV for lithiation and 170 mV and 440 mV for delithiation, respectively. The short second delithiation plateau (B2) observed in the purely galvanostatic measurements mentioned above can be anticipated at 336 mV. Plotting the potential–time curves vs. the nominal Li content (curve “0.17 C relax” in Fig. 3b) the OCP regions disappear since at OCP no external current flows. In this plot the similarity to the purely galvanostatic measurements (curve “0.17 C” in Fig. 3b) becomes more clear. The two lithiation potential plateaus and all three delithiation plateaus are present. Also the

Fig. 3. GITT-measurements. Galvanostatic periods at 0.17 C (1000 s) and OCP periods (2000 s) alternate during lithiation–delithiation cycling. (a): Evolution over time of current and potential during the GITT measurements. (b): Potential-Licontent plots of the GITT-measurements from Fig. 3a (curve “0.17 C relax”), purely galvanostatic measurements at 0.17 C from Fig. 2b (curve “0.17 C”) and GITTmeasurements with manually removed periods of potential adjust (curve “corrected”, compare text). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

intermediate plateaus (IA1, IB1) are visible. Moreover the potential values of the plateau potentials are quasi identical to the values obtained during the purely galvanostatic measurements at the same C-rate (0.17 C). However, the maximum Li content during lithiation appears with 3.4 Li equivalents to be much higher than in the purely galvanostatic measurements (2.95). In contrast, during lithiation the values are very similar with in both cases 2.54 equivalents of extracted Li. The reason for the apparent difference in Li content during lithiation are the periods after switching from OCP periods back to galvanostatic treatment. Here reformation and reorganization of the electrochemical double layer and the reestablishment of concentration gradients takes place. The current flowing during these periods is mostly not related to lithiation. After removing these periods the resulting potential–time curve is represented by the curve “0.17 C corrected” in Fig. 3b. The maximum Li content is lower now than in the purely galvanostatic measurements. Taking into account that lithiation continues already during the return of potential to the actual lithiation potential, together with reformation and reorganization of the electrochemical double layer and the reestablishment of concentration gradients, giving rise to a certain fraction of additional current in these periods, the degree of lithiation in the GITT measurements is similar to the purely galvanostatic measurements. 3.4. Potentiostatic step lithiation–delithiation cycles For comparison to the galvanostatic lithiation–delithiation cycle measurements also stepwise potentiostatic measurements were conducted (Fig. 4a and b). To avoid very large currents the

P. Bach et al. / Electrochimica Acta 164 (2015) 81–89

Fig. 4. (a): Stepwise potentiostatic lithiation–delithiation cycles employing a current limit of 1 mA as trigger. The inset represents a magnification of the time span around 0 V indicated by the frame in the main figure. (b): Comparison of potential-Li-content plots corresponding to stepwise potentiostatic lithiation– delithiation cycles employing current limits of 1 mA and 5 mA. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

potential was not just switched between the upper and lower potential limit used in the galvanostatic experiments during cycling. Instead the potential was reduced stepwise. The time limit for each step was defined by the current having dropped to a set current limit. Two different current limits (5 mA and 1 mA) were tested. In the measurement applying a current limit of 1 mA (Fig. 4a), starting from 1.5 V obviously the first potential steps are only narrow quickly reaching lower values, since no faradaic current occurs here at high potentials. The first lithiation potential plateau (A1 in Fig. 4a)) is visible in these measurements comprising the potential steps at 220 mV and 210 mV which are drastically longer than the steps before. Obviously between 220 mV and 210 mV a faradaic electrochemical process takes place giving rise to the first lithiation plateau in the potential–time curve (A1 in Fig. 4a). The following potential steps are significantly shorter. The first (faradaic) electrochemical process is completed. The second lithiation potential plateau (A2) is subsequently visible at 90 mV, 80 mV and 70 mV. After reaching the lower potential limit of 5 mV the potential is raised successively again, finally forming the first delithiation potential plateau (B1) at 170 mV, 180 mV and 190 mV and the short second delithiation plateau (B2) at 340 mV and 350 mV. A further delithiation potential plateau follows formed by the potential steps at 450 mV, 460 mV and 470 mV (B3). After this last delithiation potential plateau the potential steps return to be significantly shorter and the potential is increased quickly towards the upper potential limit of 1.5 V to keep the current above the current limit. Also some hints for the presence of intermediate plateaus observed during the galvanostatic measurements are detectable expressed by specific slightly

85

wider potential steps within the regions of quick potential change. The potential–time curve of the measurements employing the 5 mA current limit (not shown here) is similar to the one at 1 mA (described above). In Fig. 4b the potential-Li-content curve of the potentiostatic lithiation–delithiation experiments acquired employing the 1 mA and 5 mA current limit are shown in comparison. The high qualitative similarity is obvious. However some small quantitative differences are visible with slightly shifted plateau potentials at the 5 mA limit. As observed comparing the galvanostatic measurements at the different currents, a difference in nominal Li content between the measurements employing potential limits of 5 mA and 1 mA gets apparent. While at 5 mA only an amount of charge capable to reduce 2.72 Li atoms per Au atom was converted during lithiation at 1 mA 2.88 Li atoms per Au atom were nominally reduced. In both cases the amount of Li extracted during delithiation was 2.56 equivalents. In Fig. 5 the measured current–time curves during the different potential steps of a cycle at 1 mA are exemplified around the plateau potentials of the second lithiation plateau. The selected individual current–time curves are plotted normalized to the total duration of the respective potential step (ttotal), each. In general during each potential step the current drops down beginning from a start value. Apart from the potential plateaus, where the potential steps are significantly longer the current drops down quickly to the current limit. During the potential plateaus not only the potential steps are longer, also the detailed evolutions significantly differ from the usual exponential-decay-like shape. Characteristic changes are detectable during a potential plateau. In Fig. 5 at the 160 mV step the current evolution with time shows a rather exponential decay. Approaching the plateau potential (90 mV) the shape of the current-time curve is changed in a characteristic manner. The current evolution at 120 mV gets out of line since this is the potential step assigned to the intermediate plateau (IA1) known from the galvanostatic measurements. At 90 mV the current-time curve is significantly different from the current-time curve at the potential steps before. After an initial current decrease the current starts rising again, passes a maximum and decreases finally. During the further potential steps of the plateau the current evolution is still different from the current evolution observed before the potential plateau is installed, but without current dip.

Fig. 5. Current–time curves during stepwise potentiostatic lithiation–delithiation cycles employing a current limit of 1 mA as trigger. The shown curves correspond to the second lithiation potential plateau A2 in Fig. 4a. The curves are normalized to the total duration of the respective step within this plateau (ttotal) and shifted vertically for clarity. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

86

P. Bach et al. / Electrochimica Acta 164 (2015) 81–89

3.5. Further Discussion A first indication for a mechanistic difference between alloying and dealloying is provided by the CVs. While the reduction peak in the CV does not show sub-peaks, three oxidation peaks are visible which represent three related reversible dissolution steps of the formed Li–Au alloy phases. These dissolution steps point directly to three different alloy phases. The different paths of alloy formation and alloy dissolution, respectively, are further substantiated by the galvanostatic potential–time curves. During lithiation (i.e. reduction) there are only two plateaus while during delithiation (i.e. oxidation) there are three. The number of delithiation plateaus in the galvanostatic potential–time curves corresponds to the number of oxidation peaks in the anodic branch of the CV. The curves from Fig. 2b are similar to the work on Au thin film electrodes reported by Taillades [10], but there only two delithiation plateaus were distinguished. The potentiodynamic (CV) and galvanostatic approaches do not necessarily trigger the same reactions and phases. Nevertheless, there is a very good quantitative correlation between the oxidation peak positions in the CV and the plateau potentials in the galvanostatic measurements (see Fig. 1b). This suggests that the respective phase transformations during delithiation are identical in the CVs and the galvanostatic measurements. The linear correlation relates to the ohmic resistance of the cell. Nevertheless, the presence of only a single reductive peak in the CV, instead of the two expected from the galvanostatic scans can be understood due to the continuous and relatively fast change in potential during the CVs. Before a new alloy front can proceed fully through the film the potential is further lowered to form a higher lithiated phase at the surface. In result there are no distinct reduction peaks in the cathodic branch of the CVs. The Solid-Electrolyte-Interphase (SEI)-film which is forming due to an initial decomposition of electrolyte plays an important role in Li ion transfer. On graphite electrodes the SEI forms in two major stages [28]. The first stage well above the lithiation potential results in a loose and highly resistive film. The second stage occurs simultaneously with lithiation and a stable, compact and highly conductive SEI film is formed. Similar processes were reported at a silicon nanowire anode [29,30]. Also our lithiation–delithiation experiments showed a clear shoulder during the initial potential decrease (“S” in Fig. 2a). Analogue to the mentioned reports in literature this feature can be assigned to the first stage of SEI-film formation. The first stage of SEI-film formation can also be recognized in the CV measurements where the inset in Fig. 1a shows a small cathodic peak at 550 mV. The second stage of SEI formation was reported to superimpose to Li alloying. Before the first cycle the sample represents an asprepared, still flat and dense surface without any SEI-film. During the reduction peak in the first CV the full, second-stage SEI forms and the first Li atoms insert into the Au film. With a nearly reversible lithiation–delithiation process the difference between reductive and oxidative current (“Li excess” in Fig. 1c) can be attributed to this SEI-film formation. Only in the first few cycles a significant difference is visible. At higher cycle numbers the coulombic efficiency in the CVs reaches 95%. The still existing difference could be attributed to still necessary reformation of SEI for example due to dissolution or disordering of the SEI during the delithiation periods. Further SEI growth may be well triggered by an increasing roughness as well as possible crack formation in the film. Also the galvanostatic cycles show SEI-film formation. While the first stage causes the mentioned shoulder (“S” in Fig. 2a), the second step of SEI-film formation may contribute to the observed dips in potential before the first lithiation plateau in the galvanostatic curves. Furthermore, the lithiation branch in the galvanostatic measurements is in all cycles longer than the

delithiation branch while the shape of the curves is almost identical after the first cycle. This may be explained by reformation or rearrangements of the SEI during the lithiation cycles. Consequently, the total Li content of the film can be best determined from the shorter delithiation branch. The coulombic efficiency obtained in the presented galvanostatic measurements was 94% at 1.7 C and 85% at 0.17 C. Taillades et al. [10] reported 57% at 0.63 C and Yuan et al. [17] 31.5% at 0.39 C. Although the related mechanisms are not entirely clear, the SEI facilitates the insertion into the active material [28,29]. This fact can well explain the observed dip or current minimum before the galvanostatic plateaus, and can account for the shift in the onset of lithiation in the voltammograms. Nevertheless, the effect of the SEI must be in addition to an effect of the necessary initial nucleation of different subsequent phases which form with increasing Li content. Once a new phase is present after an initial nucleation its respective growth is typically facilitated, which may also explain above behavior. Especially the less pronounced dips occurring before the second plateau and at higher cycles are in line with the latter explanation. Taillades observed similar dips during lithiation of silver and gold thin films which were attributed to polarization effects [10,31]. Between the first and second cycle of the galvanostatic curve a clear difference in the potentials of the respective first lithiation plateau is present on our Au thin films (A' and A in Fig. 2a). Silicon, a promising future high-capacity anode material, shows a very similar behavior which is associated with a transition from an initially crystalline towards an amorphous state [29,32–36]. In consequence, also in our experiments on Au a similar transition is obvious which also explains the shift in the onset of lithiation in the cyclic voltammograms. During the stepwise potentiostatic lithiation–delithiation cycles shown in Figs. 4 and 5 current flows at all potential steps, but mainly at the plateaus. The shape of the current evolution at the short steps reminds to an exponential decay associated to electrochemical double layer charging and comparable to the intermittent steps during the GITT measurements presented in Fig. 3. In particular at the potential plateaus this current is superimposed by the faradaic current for Li reduction or oxidation. Considering the current evolution at 90 mV (second lithiation plateau) of Fig. 5 the lithiation current is not established instantaneously. The reason for the retarded current are obviously above discussed kinetic reasons such as the interplay between nucleation and growth, and SEI-film formation and rearrangement. While in the galvanostatic cycles the potential had to be adjusted to maintain the forced current, here the current is not constant and has hence time to develop. The initial contribution of the current due to charge or discharge of the double layer capacitor avoids a fast current drop. Together this leads to the appearance of intermediate maxima only in the current time curves at the first potential step(s) of the plateaus. At the later steps the SEI and necessary nuclei have been already formed. Such phenomena were generally observed in our potentiostatic measurements in the current evolution of the first, and to smaller extend also of the second, potential step of both lithiation plateaus (see above). Slightly different shapes of current time-curves at individual step before and after a specific potential plateau (see Fig. 5) are in line with differences in the structure of the electrochemical double layer at the different alloy phase surfaces. Fig. 6 shows a comparison between the potential–time curves measured galvanostatically at 1.7 C (16.6 mA) and 0.17 C (1.66 mA) and by stepwise potentiostatic scans employing current limits of 5 mA and 1 mA. The respective curves of the two modes show high qualitative similarity. The two main potential plateaus during lithiation (A1, A2) and the three potential plateaus during delithiation (B1, B2, B3) are clearly recognizable. The intermediate

P. Bach et al. / Electrochimica Acta 164 (2015) 81–89

87

Table 1 Li content of Li–Au alloy phases. Li content x in LixAu

Lithiation state 1 (L1) Lithiation state 2 (L2) Delithiation state 1 (D1) Delithiation state 2 (D2) Delithiation state 3 (D3) Delithiation state 4 (D4)

Fig. 6. Comparison between galvanostatic measurements at currents of 1.7 C (16.6 mA) and 0.17 C (1.66 mA) and stepwise potentiostatic measurements employing current limits of 5 mA and 1 mA. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

plateaus observed in the galvanostatic measurements are also detectable in the stepwise measurements by wider potential steps. Also the maximum nominal Li content during delithiation was found to be almost identical for both modes. Employing a current limit of 1 mA 2.56 equivalents were nominally extracted, employing a current limit of 5 mA 2.53, and for the galvanostatic cycles at 0.17 C 2.54, respectively. The numbers point to Li3Au as the highest lithiated phase in our experiments, which is confirmed by an ongoing in-situ XRD study to be reported in a separate publication. The difference in nominal Li content between 2.55 and 3 may be a consequence of volume expansion cycles. Part of the initial Li may be not accessible anymore considering loss of electrical contact of parts of the initial Au film. Another possible explanation is incomplete dissolution of the alloy phases leaving a Li-poor alloy phase instead of recovered Au. At the higher rate of 1.7 C the maximum obtained Li content is obviously smaller. With higher rates additional limitations related to diffusion phenomena in the electrode material are expected. The very good correspondence between the galvanostatic and potentiostatic experiments reveals the same reaction paths followed during lithiation and delithiation within the covered current range. In order to illuminate the structure formation of the Li–Au binary system, we have performed a theoretical structure research, using the minima hopping method (MHM) [37,38] as the global structure predictor. By coupling with ab-initio density functional theory (DFT), this method is capable of predicting stable and metastable crystal structures at a given pressure, from the sole knowledge of the chemical composition of the system. This methodology has shown remarkable results in the prediction of novel structures in a wide range of materials [39–41]. We have built the phase diagram for Li–Au, where forty three different stoichiometries starting from pure Li to pure Au were studied. We found that this system has a large number of structures with negative enthalpy, much larger than similar systems which were addressed in parallel, such as LiAl [42] or LiSi [43]. This directly points to a very large number of meta-stable structures as compared with Li–A alloys, where A is an atom with p-valence electrons. Among others we obtained structures of different stoichiometries such as LiAu2 (monoclinic, P121/m1), LiAu (monoclinic, P12/m1), Li5Au3 (rhombohedral, R32), or Li3Au (cubic, Fm – 3 m). A complete report will be presented elsewhere together with a full description of the obtained structural phases. Our preliminary analysis of the structural simulations points to a high number of (meta) stable phases expected for the Li–Au system (for example compared to Li–Al or Li–Si) which is well in line with

Model 1: remaining Li

Model 2: contact loss

1.60 3.00 1.60 1.49 0.71 0.58

1.27 3.00 1.27 1.15 0.22 0.075

the analysis of the electrochemical experiments presented here and our experimental results for the Li–Al system [44]. In Table 1 we present alloy phases together with their Li content calculated from the galvanostatic measurement at 0.17 C. The maximum nominal Li content during delithiation was assumed to be x = 3.0 equivalents. Depending on which of the above discussed arguments for the extracted 2.55 equivalents of Li during delithiation applies, two different Li contents can be calculated. For Li remaining inside the sample after the first delithiation the Li concentrations are higher (Model 1) than if less electrode material is accessible after the first delithiation due to contact loss during the first cycle (Model 2). A summary of the mechanistic sequence is represented in Figs. 7 and 8. Fig. 7 shows a sketch of the lithiation and delithiation process. Fig. 8 marks the alloy phase evolution stages for the first (Fig. 8a) and the subsequent cycles (Fig. 8b) of the Au–Li system. The colored bars indicate the single-phase regions of the denoted phases where the major potential changes occur between the respective plateaus. The single-phase region of the respective starting phase of the different cycles is only indicated starting after the SEI is formed, i.e. after the initial potential dip. The first lithiation plateau at 160 mV (after the SEI-film formation in the first lithiation half-cycle) is assigned to a two phase region between pristine, crystalline Au and a Li–Au alloy phase, (lithiation state L1). Then follows a region of decreasing potential, which is indicative for a single phase region. Here only state L1 is present. The following second plateau at 60 mV represents a two phase

Fig. 7. Mechanistic model of the lithiation–delithiation cycles at a 100 nm thin Au model electrode including pristine Au and the lithiation states L1 and L2, and delithiation states D1 to D4. The phase transformations A1, A2, B1–B3 and IB3 are analogue to Fig. 2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

88

P. Bach et al. / Electrochimica Acta 164 (2015) 81–89

cycles. This is supported by the observation of identical potential values for the second plateau in the initial and second cycle. 4. Conclusion During the galvanostatic and potentiostatic lithiation–delithiation cycles two main potential plateaus during lithiation and three potential plateaus during delithiation are visible. In addition intermediate plateaus are present. The potential dips at the beginning of the lithiation plateaus in the galvanostatic measurements (D1, D2 in Fig. 2) have an analogy in the current–time curves of the stepwise potentiostatic measurements by a retarded raise of current. The observations can be explained by SEI-film formation as well as by nucleation phenomena during subsequent phase formations. The amount of Li extracted during delithiation was equal in the potentiostatic and galvanostatic measurements. Lithiation results in different phases leading to a final Li3Au phase. The results presented point to the same reaction path to be followed during lithiation and delithiation in both types of measurements within the covered current range. In particular the appearance of the same alloy phases and the presence of the same phase transformations are evidenced by our different electrochemical measurements. Based on the presented results a mechanistic model and an alloy phase evolution diagram for the lithiation–delithiation cycles of the Au film model electrode could be established including five different Li–Au-alloy phases (L1, L2, D1–D3) and a Li-poor boundary phase (D4). Acknowledgement

Fig. 8. Alloy phase evolution diagrams of the first (a) and subsequent (b) electrochemical lithiation–delithiation cycles of a 100 nm thin Au film model electrode. The lithiation states are labeled L1 and L2, whereas the delithiation states D1–D4. The colored bars indicate the single-phase regions of the denoted phases. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

P. B. acknowledges financial support from the European Union (EU) within the framework of the Center for Electrochemical Sciences (CES) at the Ruhr University Bochum, Germany. This work was partially supported by the German Federal Ministry for Education and Research (BMBF) in the framework of the Kompetenzverbund Nord project. References

region between state L1 and state L2. At the end of lithiation only state L2 is present as it can be seen from the decreasing potential. From the galvanostatic measurements and the related transferred charges the L2 state can be associated with the Li3Au phase. We did not observe any further transition to a higher lithiated state such as the Li15Au4 phase. The first delithiation plateau at 195 mV follows, denoting a two phase region between lithiation state L2 and delithiation state D1. After the subsequent increase of potential (single phase region of D1) the small second pronounced plateau at 380 mV is reached where state D1 transforms into state D2. During the single phase region of state D2 the potential increases to reach the last delithiation plateau at 480 mV where state D2 and state D3 coexist. First results of a currently ongoing in-situ XRD study suggest correlating the intermediate plateau observed within the final potential increase (IB3) to a further phase transformation leading to delithiation state D4. The results from the galvanostatic measurements (and XRD results not shown here) suggest that D4 is not (crystalline) Au as discussed above. D4 is probably a Li-poor phase since it is the ending-phase of delithiation and pure (lithium free) Au is not recovered. During subsequent second Li loading the state D4 (instead of the initial pure Au) forms a two phase region with a Li–Au alloy phase, similar to the initial cycle. The single phase region of the state D4 is thus followed by a plateau, which is however at a different potential than the first lithiation plateau in the initial cycle since D4 now is lithiated but not Au. The alloy phase formed during the first lithiation plateaus is nevertheless the same in both

[1] D.M. Kolb, Structure studies of metal electrodes by in-situ scanning tunneling microscopy, Electrochim. Acta 45 (2000) 2387. [2] F. Schreiber, Structure and growth of self-assembling monolayers, Prog. Surf. Sci. 65 (2000) 151. [3] Th. Wink, S.J. van Zuilen, A. Bult, W.P. van Bennekom, Self-assembled monolayers for biosensors, Analyst (1997) 43R. [4] D.O. Demirkol, H.B. Yildiz, S. Sayınc, M. Yilmazc, Enzyme immobilization in biosensor constructions: self-assembled monolayers of calixarenes containing thiols, RSC Adv. 4 (2014) 19900. [5] D.D. Liana, B. Raguse, L. Wieczorek, G.R. Baxter, K. Chuah, J.J. Gooding, E. Chow, Sintered gold nanoparticles as an electrode material for paper-based electrochemical sensors, RSC Adv. 3 (2013) 8683. [6] A. Ahmad, E. Moore, Electrochemical immunosensor modified with selfassembled monolayer of 11-mercaptoundecanoic acid on gold electrodes for detection of benzo[a]pyrene in water, Analyst 137 (2012) 5839. [7] T. Kondo, J. Morita, K. Hanaoka, S. Takakusagi, K. Tamura, M. Takahasi, J. Mizuki, K. Uosaki, Structure of Au(111) and Au(1 0 0) single-crystal electrode surfaces at various potentials in sulfuric acid solution determined by in-situ surface Xray scattering, J. Phys. Chem. C 111 (2007) 13197. [8] M. Thakur, M. Isaacson, S.L. Sinsabaugh, M.S. Wong, S.L. Biswal, Gold-coated porous silicon film as anodes for lithium ion batteries, J. Power Sources 205 (2012) 426. [9] Y. Yu, L. Gu, X. Lang, C. Zhu, T. Fujita, M.W. Chen, J. Maier, Li Storage in 3D nanoporous Au-supported nanocrystalline tin, Adv. Mater. 23 (2011) 2443. [10] G. Taillades, N. Nejelloun, J. Sarradin, M. Ribes, Metal-based very thin film anodes for lithium ion microbatteries, Solid State Ionics 152–153 (2002) 119. [11] F.U. Renner, H. Kageyama, Z. Siroma, M. Shikano, S. Schröder, Y. Gründer, O. Sakata, Gold model anodes for Li-ion batteries: Single crystalline systems studied by in situ X-ray diffraction, Electrochim. Acta 53 (2008) 6064. [12] S. Misra, N. Liu, J. Nelson, S.S. Hong, Y. Cui, M.F. Toney, In situ X-ray diffraction studies of (de) lithiation mechanism in silicon nanowire anodes, ACS Nano 6 (2012) 5465. [13] D. Borissov, A. Pareek, F.U. Renner, M. Rohwerder, Electrodeposition of Zn and Au–Zn alloys at low temperature in an ionic liquid, Phys. Chem. Chem. Phys. 12 (2010) 2059.

P. Bach et al. / Electrochimica Acta 164 (2015) 81–89 [14] K. Nishio, K. Yuda, H. Masuda, Fabrication of nanoporous gold film by electrochemical alloying and dealloying with Li at room temperature, ECS Electrochem. Lett. 2 (1) (2013) C1. [15] M., Raney, 1927. Method of producing finely divided nickel, U.S. Patent 1628190. [16] M. Oezaslan, F. Hasché, P. Strasser, Pt-based core–shell catalyst architectures for oxygen fuel cell electrodes, J. Phys. Chem. Lett. 4 (19) (2013) 3273. [17] L. Yuan, H.K. Liu, A. Maaroof, K. Konstantinov, J. Liu, M. Cortie, Mesoporous gold as anode material for lithium-ion cells, J. New Mat. Elect. Syst. 10 (2007) 95. [18] O.M. Magnussen, Ordered anion adlayers on metal electrode surfaces, Chem. Rev. 102 (3) (2002) 679. [19] Y. Nakai, M.S. Zei, D.M. Kolb, G. Lehmpfuhl, A LEED and RHEED investigation of Cu on Au(111) in the underpotential region, Ber. Bunsen Phys. Chem. 88 (1984) 340. [20] M.F. Toney, J.N. Howard, J. Richer, G.L. Borges, J.G. Gordon, O.R. Melroy, D. Yee, L. B. Sorensen, Electrochemical deposition of copper on a gold electrode in sulfuric acid: resolution of the interfacial structure, Phys. Rev. Lett. 75 (1995) 4472. [21] O.M. Magnussen, J. Hotlos, G. Bettel, D.M. Kolb, R.J. Behm, Atomic Structure of ordered copper adlayers on single-crystalline gold electrodes, J. Vac. Sci. Technol. B 9 (2) (1991) 969. [22] S. Virtanen, P. Schmuki, H.S. Isaacs, In situ X-ray absorption near edge structure studies of mechanisms of passivity, Electrochim. Acta 47 (19) (2002) 3117. [23] B. Ocko, J. Wang, A. Davenport, H. Isaacs, In situ x-ray reflectivity and diffraction studies of the Au(001) reconstruction in an electrochemical cell, Phys. Rev. Lett. 65 (1990) 1466. [24] O.R. Melroy, M.F. Toney, G.L. Borges, M.G. Samant, J.B. Kortright, P.N. Ross, L. Blum, An in-situ grazing incidence X-ray scattering study of the initial stages of electrochemical growth of lead on silver (111), J. Electroanal. Chem. 258 (1989) 403. [25] J. Zegenhagen, T.-L. Lee, Y. Gruender, F.U. Renner, B.O. Fimland, The adsorption and growth of copper on As-terminated GaAs(001): Physical vapour versus electrochemical deposition, Z. Phys. Chem. 221 (2007) 1273. [26] W. Weppner, R.A. Huggins, Determination of the Kinetic Parameters of Mixed Conducting Electrodes and Application to the System Li3Sb, J. Electrochem. Soc. 124 (10) (1977) 1569. [27] F. Alessandrini, G.B. Appetecchi, M. Conte, S. Passerini, Ionic liquid based electrolytes for high energy electrochemical storage devices, ECS Trans. 200 (2015) 67. [28] S. Zhang, M.S. Ding, K. Xu, J. Allen, T.R. Jow, Understanding solid electrolyte interface film formation on graphite electrodes, Electrochem. Solid St. Lett. 4 (12) (2001) A206.

89

[29] C.K. Chan, R. Ruffo, S.S. Hong, R.A. Huggings, Y. Cui, Structural and electrochemical study of the reaction of lithium with silicon nanowires, J. Power Sources 189 (2009) 34. [30] J. Graetz, C.C. Ahn, R. Yazami, B. Fultz, High reversible lithium storage in nanostructured silicon, Electrochem. Solid St. Lett. 6 (2003) A194. [31] G. Taillades, J. Sarradin, Silver: High performance anode for thin film lithium ion batteries, J. Power Sources 125 (2004) 199–205. [32] A. Netz, R.A. Huggins, W. Weppner, The formation and properties of amorphous silicon as negative electrode reactant in lithium systems, J. Power Sources 119–121 (2003) 95. [33] M.N. Obrovac, L.J. Krause, Reversible Cycling of Crystalline Silicon Powder, J. Electrochem. Soc. 154 (2) (2007) A103. [34] T.D. Hatchard, J.R. Dahn, In-situ XRD and electrochemical study of the reaction of Li with amorphous silicon, J. Electrochem. Soc. 151 (6) (2004) A838. [35] J. Li, J.R. Dahn, An in situ X-ray diffration study on the reaction of Li with crystalline Si, J. Electrochem. Soc. 154 (3) (2007) A156. [36] C.K. Chan, H. Peng, G. Liu, K. McIlwrath, X.F. Zhang, R.A. Huggins, Y. Cui, High performance lithium battery anodes using silicon nanowires, Nature nanotech. 3 (2008) 31. [37] S. Goedecker, Minima hopping: An efficient search method for the global minimum of the potential energy surface of complex molecular systems, J. Chem. Phys. 120 (2004) 9911. [38] M. Amsler, S. Goedecker, Crystal structure prediction using the minima hopping method, J. Chem. Phys. 133 (2010) 224104. [39] R. Sarmiento-Pérez, T.F.T. Cerqueira, I. Valencia-Jaime, M. Amsler, S. Goedecker, S. Botti, M.A.L. Marques, A.H. Romero, Sodium–gold binaries: novel structures for ionic compounds from an ab initio structural search, New J. Phys. 15 (2013) 115007. [40] J.A. Flores-Livas, M. Amsler, T.J. Lenosky, L. Lehtovaara, S. Botti, M.A.L. Marques, S. Goedecker, High-Pressure Structures of Disilane and Their Superconducting Properties, Phys. Rev. Lett. 108 (2012) 117004. [41] M. Amsler, J.A. Flores-Livas, T.D. Huan, S. Botti, M.A.L. Marques, S. Goedecker, Novel Structural Motifs in Low Energy Phases of LiAlH4, Phys. Rev. Lett. 1082 (2012) 205505. [42] T.F.T. Sarmiento-Pérez, I. Cerqueira, M. Valencia-Jaime, S. Amsler, A.H. Goedecker, S. Romero, M.A.L. Botti, J. Chem. Phys. 142 (2015) 024710. [43] I. Valencia-Jaime, R. Sarmiento-Pérez, S., Botti, M.A. L. Marques, M., Amsler, S., Goedecker, A.H. Romeo, submitted. [44] P. Bach, A. Seemayer, U. Rütt, O. Gutowski, F.U. Renner, Electrochemical Cells for In-Situ XRD Studies of Insertion and Extraction Mechanisms of Lithium in Anode Materials for Lithium Ion Batteries Tested at Aluminum Model Electrodes, ECS Transactions 53 (29) (2013) 29–39.