In Situ Real-Time Mechanical and Morphological Characterization of

Jan 3, 2018 - ... Characterization of Electrodes for Electrochemical Energy Storage ... small/large particles, binders with different mechanical prope...
2 downloads 3 Views 8MB Size
Article Cite This: Acc. Chem. Res. 2018, 51, 69−79

pubs.acs.org/accounts

In Situ Real-Time Mechanical and Morphological Characterization of Electrodes for Electrochemical Energy Storage and Conversion by Electrochemical Quartz Crystal Microbalance with Dissipation Monitoring Published as part of the Accounts of Chemical Research special issue “Energy Storage: Complexities Among Materials and Interfaces at Multiple Length Scales”. Netanel Shpigel, Mikhael D. Levi,* Sergey Sigalov, Leonid Daikhin, and Doron Aurbach* Department of Chemistry and BIU Center for Nanotechnology and Advanced Materials (BINA), Bar-Ilan University, Ramat Gan 5290002, Israel CONSPECTUS: Quartz crystal microbalance with dissipation monitoring (QCM-D) generates surface-acoustic waves in quartz crystal plates that can effectively probe the structure of films, particulate composite electrodes of complex geometry rigidly attached to quartz crystal surface on one side and contacting a gas or liquid phase on the other side. The output QCM-D characteristics consist of the resonance frequency (MHz frequency range) and resonance bandwidth measured with extra-ordinary precision of a few tenths of Hz. Depending on the electrodes stiffness/ softness, QCM-D operates either as a gravimetric or complex mechanical probe of their intrinsic structure. For at least 20 years, QCM-D has been successfully used in biochemical and environmental science and technology for its ability to probe the structure of soft solvated interfaces. Practical battery and supercapacitor electrodes appear frequently as porous solids with their stiffness changing due to interactions with electrolyte solutions or as a result of ion intercalation/adsorption and long-term electrode cycling. Unfortunately, most QCM measurements with electrochemical systems are carried out based on a single (fundamental) frequency and, as such, provided that the resonance bandwidth remains constant, are suitable for only gravimetric sensing. The multiharmonic measurements have been carried out mainly on conducting/redox polymer films rather than on typical composite battery/supercapacitor electrodes. Here, we summarize the most recent publications devoted to the development of electrochemical QCM-D (EQCM-D)-based methodology for systematic characterization of mechanical properties of operating battery/supercapacitor electrodes. By varying the electrodes’ composition and structure (thin/thick layers, small/large particles, binders with different mechanical properties, etc.), nature of the electrolyte solutions and charging/cycling conditions, the method is shown to be operated in different application modes. A variety of useful electrode-material properties are assessed noninvasively, in situ, and in real time frames of ion intercalation into the electrodes of interest. A detailed algorithm for the mechanical characterization of battery electrodes kept in the gas phase and immersed into the electrolyte solutions has been developed for fast recognition of stiff and viscoelastic materials in terms of EQCM-D signatures treated by the hydrodynamic and viscoelastic models. Working examples of the use of in situ hydrodynamic spectroscopy to characterize stiff rough/porous solids of complex geometry and viscoelastic characterization of soft electrodes are presented. The most demonstrative example relates to the formation of solid electrolyte interphase on Li4Ti5O12 electrodes in the presence of different electrolyte solutions and additives: only a few cycles (an experiment during ∼30 min) were required for screening the electrolyte systems for their ability to form high-quality surface films in experimental EQCM-D cells as compared to 100 cycles (200 h cycling) in conventional coin cells. Thin/small-mass electrodes required for the EQCM-D analysis enable accelerated cycling tests for ultrafast mechanical characterization of these electrodes in different electrolyte solutions. Hence, this methodology can be easily implemented as a highly effective in situ analytical tool in the field of energy storage and conversion.



INTRODUCTION

The use of electrochemical energy storage in both large-scale

Received: September 27, 2017 Published: January 3, 2018

application devices and portable electronics requires the © 2018 American Chemical Society

69

DOI: 10.1021/acs.accounts.7b00477 Acc. Chem. Res. 2018, 51, 69−79

Article

Accounts of Chemical Research

Figure 1. Gravimetric and nongravimetric applications of EQCM-D for the characterization of energy-storage electrodes. (bottom panel) Acoustic waves for fundamental frequency and its 3rd overtone.

coating shifts the values of f and W from that typical for the pristine (uncoated) QC. Immersion of coated crystal into a liquid causes additional changes in f and W. The entire shifts of resonance frequency and resonance width (Δf and ΔW, respectively) contain encoded information about the electrode material properties. In particular, the following typical operation modes of EQCM-D can be useful for electrode characterization: (i) For thin, stiff, nonporous solid electrodes with flat external surfaces, EQCM operates as a gravimetric sensor as long as there are no changes in the resonance width (i.e., ΔW = 0). Any change in the frequency caused by either adsorption of species from the gas phase or their electrochemical adsorption from electrolyte solutions can be directly converted to the related mass change using the Sauerbrey equation.6,8 (ii) By contrast, any deviation from an ideal flat surface morphology and/or the appearance of a stiff porous electrode layer leads to additional frequency changes coupled with a significant energy dissipation (i.e., ΔW > 0) due to the hydrodynamic solid−liquid interactions. (iii) Finally, if the electrode coating is soft (i.e., can be represented by a viscoelastic plate rigidly attached to the QC surface6), the changes in Δf/n and ΔW/n do occur but depend on the overtone order in an entirely different functional form compared to case (ii). Although gravimetric information can be obtained with the use of a single oscillation frequency (when ΔW = 0), in situ mechanical and morphological quantification of the electrode properties requires measurements of Δf/n and ΔW/n for several overtone orders using either a multiharmonic network analyzer9 or EQCM-D.6,7,10 The number of measured overtone responses should be greater or equal to the number of structural parameters retrieved by the fitting routines using equations describing hydrodynamic or viscoelastic interactions. A proper choice of thin electrode preparation and charging/ cycling conditions enables the use of EQCM-D as a unique in situ gravimetric, hydrodynamic (morphological), and viscoelastic probe of battery and supercapacitor electrodes as is shown in the top panel of Figure 1.

development of advanced electrode materials possessing both high energy and high power densities during the long-time operation of these devices. For a better understanding of the underlying intercalation/adsorption mechanisms to be acquired, many structure-sensitive surface spectroscopy/microscopy techniques have been proposed.1 These techniques include high-resolution methods designed to probe the local structure of electrode hosts as well as those that characterize their macroscopic mechanical properties. However, most of these techniques are capable of exploring only a single material property so that a combination of complementary methods is usually required to fully understand the behavior of the electrode. In this Account, we present a conceptually different approach that allows continuous monitoring of the gravimetric, dimensional, and viscoelastic changes in battery electrodes. By varying the composition and geometric parameters of the tested electrodes, such as thin/thick coatings, small/large particles, different nature of binders and electrolyte solutions, and finally the charging/cycling conditions, a variety of gravimetric and mechanical characteristics of the operating electrodes can be assessed using an electrochemical quartz crystal microbalance (EQCM) with dissipation monitoring (EQCM-D). This method provides noninvasive in situ realtime monitoring of these properties during long-term cycling of batteries and supercapacitor electrodes. The two output channels of EQCM-D, namely resonance frequency and dissipation, make it highly suitable for both gravimetric and combined gravimetric and mechanical characterization of thin battery and supercapacitor electrodes.2−7 The dissipation factor D is defined as the ratio of the full-width at half-height of the resonance peak (i.e., bandwidth W) to the resonance frequency, f: D = W/f.6 For brevity, we will refer to W as the “resonance width”. As can be seen in Figure 1 (bottom), the acoustic shear waves generated by a piezoelectric crystal, operating in horizontal-shear (or thickness-shear) mode, penetrate a thin electrode coating rigidly attached to a quartz crystal (QC) surface in contact with a gas or liquid (here “rigidly attached” means no-slipping condition at the interface).2−7 The loading 70

DOI: 10.1021/acs.accounts.7b00477 Acc. Chem. Res. 2018, 51, 69−79

Article

Accounts of Chemical Research



BRIEF BACKGROUND OF EQCM

case (i) is suitable only for ex situ characterization of the hydrodynamic solid−liquid interactions (because these liquids may not be ionically conductive),16 the measurements with multiple harmonics (case (ii)) can be performed in conductive electrolyte solutions at different applied potentials. This enables in situ hydrodynamic spectroscopy of electrochemical systems, which probes the rough/porous structure of solid electrodes’ active mass of complex morphology.17 For diluted aqueous solutions, the values of δ for the fundamental frequency and for the 13th overtone are approximately equal to 240 and 68nm, respectively. This is a typically mesoscopic range of the penetration depths of the shear acoustic waves probing the rough/porous structure of the electrodes’ coatings.

Inertial and Viscous Loads

In 1959, Günter Sauerbrey found a linear relationship between the change in frequency (Δf) of a QC and the corresponding load mass (Δm). As long as the mass-induced change in the frequency is small compared to the fundamental frequency, f 0 (i.e., Δf ≪ f 0,8 called the small-load approximation6) and the coating is stiff and uniform, the Sauerbrey equation is valid: Δm = −C Δf /n

where C =

(1)

μq ρq /2f02 is the mass sensitivity constant, and μq

and ρq are the elastic shear modulus and density, respectively, for AT-cut QC. For a 5 MHz crystal C = 17.7 ng cm−2 Hz−1. Eq 1 generalizes the original Sauerbrey equation for various overtone orders, n (odd numbers from 1 to 13 for EQCM-D instrument). In the following, the change in frequency will always be designated as Δf/n. Although the Sauerbrey equation was originally found to be valid for measurements in the gas phase, it was later recognized that rigid ideally flat coatings on QC immersed in liquid also follow this equation. This discovery paved the way for the wide use of EQCM in electroanalytical and biological chemistry, environmental science, and a variety of sensors applications.6,11−13 The validity of the Sauerbrey equation for in situ EQCM measurements of electrodes in electrolyte solutions can be rationalized due to the effect of viscous solid−liquid interaction described by Kanazawa and Gordon.14 It appears that the changes in Δf/n and ΔW/n of the QC in air due to a deposited stiff flat coating is the same as measured in liquids.15 According to the Kanazawa equation, the changes in Δf/n and ΔW/n are proportional to ρL ηL , where ρL and ηL are density and dynamic viscosity of liquid, respectively. Eq 2 is written in a form that can include multiharmonic measurements (i.e., valid for different overtone orders, n). In addition, the original variable ρL ηL is substituted by the natural independent variable of the hydrodynamic problem,4,16 which is the penetration depth of the shear wave, δ (defined in eq 3).4,6,16 Further normalization of Δf and ΔW by a constant factor ρL f 20 provides a convenient correlation between the change in the EQCM-D response of the QC in contact with liquid and δ Δf nρL f02

=−

ΔW = 2nρL f02

Rigid Rough/Porous Solids

During the development of the EQCM methodologies (since 1985), it was realized that electrodes’ roughnesses should significantly affect Δf/n and ΔW/n changes.11 A consistent quantitative description of the effect of the electrodes’ roughness on the related EQCM characteristics appeared only a decade later.18 In contrast to the electrodes with flat surfaces (the hydrodynamic interactions with liquid are described by the Navier−Stokes equation), a pronounced roughness of the electrodes’ surfaces results in additional resistive Darcy-type contribution to the velocity distribution of liquid.18,19 This contribution is proportional to ηL/ξ2,18,19 where ξ2 is an important parameter of a porous electrode structure called specific permeability, which characterizes the liquid flow through the porous medium. The equations containing a complete set of the related structural parameters are solved analytically, and these analytical solutions are presented as changes in Δf/n and ΔW/n as a function of δ.4,15,16 The typical structural parameters of rough/porous solid electrodes are their density ρ or thickness l, permeability length ξ (for a uniform porous layer model), particles radius r, and coverage density of the bumps (for a surface asperities model).16 EQCM-D signatures of the rough/porous electrode structures (i.e., the characteristic coupled changes in Δf/n and ΔW/n as a function of the penetration depth δ) are presented in the next section. It is important to note that, for the EQCM-D response related to any thin stiff electrode coating, the viscous contribution to the Δf/n change (due to interaction with liquid) is additive to the inertial contribution owing to the rigidly attached mass.6 For this reason, the experimental Δf/n changes are corrected for the frequency change related to the “dry” mass of the electrode coating measured in air (Δf/nmass). The remaining Δf/n contribution together with ΔW/n changes should be fitted by a suitable hydrodynamic model.16,17

δ μq ρq δ μq ρq

Soft (Viscoelastic) Coatings

Because viscoelastic films tend to move at different velocities compared to those of the oscillating QC surface, the frequency-mass change linearity is disrupted. Early EQCM investigations of electrodeposited Ni(OH)2 layer recognized a substantial change in the motional resistance (roughly proportional to ΔW/n, n = 1)20 such that a certain deviation of the experimental Δf changes from those calculated from the charge during the film formation was detected. The result was interpreted by the fact that the electrodeposited Ni(OH)2 film is viscoelastic in nature. Only a rough estimation of the components of the complex shear modulus can be made by using Δf and ΔW when only the fundamental frequency is involved because the minimal number of model parameters is

(2)

The penetration depth δ characterizes the decay length of the shear wave across the boundary layer between the crystal and the contacting liquid (see bottom of Figure 1) δ = (hL /πnf0 ρL )1/2

(3)

Eq 3 shows that δ can be changed in two different ways: (i) either keeping the same oscillation frequency (e.g., the fundamental one) but varying the viscosity-to-density ratio (i.e., using physically different liquids) or (ii) choosing the same liquid (e.g., electrolyte solution of interest) but performing measurements on different overtones. Although 71

DOI: 10.1021/acs.accounts.7b00477 Acc. Chem. Res. 2018, 51, 69−79

Article

Accounts of Chemical Research three, whereas the number of variables resulting from the measurements is two (i.e., Δf and ΔW if only the fundamental frequency is involved). However, in the Voigt-type viscoelastic models applied to the coatings in contact with air and immersed in liquids, the Δf and ΔW changes are expressed by the different dependences as a function of n (because liquid exerts stress on the coating in contrast to that of the air phase).6,21 There are three typical fitted parameters for the conventional viscoelastic models: (i) the mass per unit area m, which is the product of the layer density ρ and thickness l (i.e., density and thickness of the coating cannot be simultaneously assessed by EQCM-D, and additional information from the complementary techniques is required, e.g., direct morphological studies by in situ AFM) and (ii) storage G′ and (iii) loss G″ shear moduli. G′ and G″ are the components of the complex shear modulus G* (these terms are well explained in the QCM-related literature). In some EQCM-D treatments, the shear viscosity of coatings, proportional to G″, is used: η = G″/ω.6 In a simple Voigt-type model, G′ and η are frequency independent. In extended viscoelastic models, the frequency dependence of both these quantities is found to be an exponential one. The exponents range is from −2 to 0 and from 0 to 2 for η and G′, respectively.6 It is important to note that, in contrast to measurements of thin stiff coatings by EQCM-D, a significant damping of the crystal oscillations across the electrodes occurs in the case of viscoelastic (soft) coatings. This leads to violation of the simple additivity rule for the frequency changes due to the “dry” mass of the coating in air and the related viscoelastic contributions. Hence, the “dry” mass of the coating must be measured in a separate experiment by a complementary technique, or alternatively, its contribution to the total frequency change should appear as one of the fitting parameters of the viscoelastic model.6 As a result, the overtone order n rather than the penetration depth δ plays the role of the independent variable of the viscoelastic model. Many prominent research teams have contributed to the viscoelastic characterization of soft electrodes. We are able in this short Account to note only a few of them.12,20,22−25

Figure 2. Gravimetric (ΔW/n = 0) and nongravimetric (ΔW/n ≠ 0) behavior of the electrodes tested by EQCM-D: schematic representation (a) and the sequence of measurement steps and data treatment for the nongravimetric case (b).

the electrode coatings. Figure 3 shows sketches of the different morphologies of stiff porous electrodes (left side). The analytical expressions for Δf/n and ΔW/n changes are collected in the appendix of ref 15. Importantly, the right side of Figure 3 reproduces the characteristic couplings of the plots of Δf/n and ΔW/n as a function of δ, which we call the EQCM-D signatures of the rough/porous electrode structure. For example, case iv (semispherical bumps) is characterized by a smaller deviation of ΔW/n as compared to that of Δf/n from the related Kanazawa straight lines (typical for plane surface). This is because the resonance width is proportional to the surface of the bumps (reflecting friction between the surface and contacting liquid), whereas the frequency change is proportional to the volume of the bumps (representing inertial load due to pushing the contacting liquid by solid bumps during oscillations).15,16 Fitting the hydrodynamic equations for Δf/n and ΔW/n to the experimental data allows the recovery of the structural parameters of the electrodes in both its initial state after immersion into the electrolyte solution and during the electrode polarization.

The Road Map: How to Choose Appropriate EQCM-D Models

The algorithm of simple tests for stiffness/viscoelasticity of the electrodes is presented schematically in Figure 2. Dealing with electrochemical intercalation/adsorption of ions, the Sauerbrey case is recognized by a finite change in Δf/n independent of the overtone order along with zero change in the resonance width ΔW/n = 0 (right side of Figure 2a). In the opposite case (i.e., when Δf/n depends on n, and ΔW/n ≠ 0) one can suspect either hydrodynamic behavior of a stiff porous solid or viscoelastic behavior of a soft electrode coating. The two cases can be distinguished by a series of EQCM-D measurements in air after immersion of the coated crystals into electrolyte solutions under an open-circuit potential and then under an applied potential (Figure 2b). The order of the required measurements, the choice of the reference state, and the characteristic plots of Δf/n and ΔW/n as functions of δ or n for the hydrodynamic and viscoelastic modeling are clearly indicated in Figure 2b (examples of practical use are presented in the following section). The initial selection of a suitable hydrodynamic or viscoelastic model to be used for fitting the experimental data is based on examination of high-resolution SEM images of 72

DOI: 10.1021/acs.accounts.7b00477 Acc. Chem. Res. 2018, 51, 69−79

Article

Accounts of Chemical Research

Figure 3. Sketches of the typical morphologies of composite electrodes (left) with the characteristic plots of the complex frequency shift (right: dashed lines). Solid lines represent the response of a flat surface. Adapted with permission from ref 15. Copyright 2017, Elsevier.



WORKED EXAMPLES

in Figure 3). When three electrodes of significantly different morphologies are immersed from air into solutions, the changes in the experimentally measured Δf/n and ΔW/n fit nicely to the related hydrodynamic models (symbols and solid lines in Figure 5b, respectively; for details of modeling and retrieving structural characteristics, see the original paper26). EQCM-D does not detect the difference in morphology of the neat Au-covered QC and Cu sputtered in a vacuum when these electrodes are immersed in liquid. The characteristic size of heterogeneities of these electrodes is below the penetration depth δ (68 nm for the highest 13th overtone) and hence is not seen by QCM-D, which gives a mesoscopic resolution.

In Situ Assessment of Complex Morphologies of Copper Deposits

Figures 4 and 5 show how EQCM-D measurements on multiple harmonics reflect the different morphologies of the electrodeposited Cu layers under various deposition conditions visualizing the hydrodynamic correction to the Sauerbrey equation. 26 During electrodeposition of Cu in cyclic voltammetric (CV) mode, more negative electrodeposition potential resulted in the larger (and thicker) coating accompanied by some dispersion of frequencies on the different overtone orders with a simultaneous increase in ΔW/n (Figure 4a). However, during a chronoamperometric (CA) Cu deposition, the frequency dispersion with n and the increase in ΔW/n with time are considerably smaller (Figure 4b). As a result, deviation of the treated data for Δf/n, as a function of the deposition charge from the Faradaic straight line (obtained by the combination of the Sauerbrey equation with the Faraday law) is significantly larger for the CVdeposited film than for the CA-formed film (Figure 4c and d, respectively). Figure 5a shows the reason for the observed dependence of the hydrodynamic effect on the electrochemical mode of Cu deposition: a uniform morphology is typical for a vacuum-sputtered Cu film. CA-deposited films have the morphology defined here as a uniform porous layer (case ii in Figure 3), whereas CV-based deposition produces micrometer-sized bumps on top of the uniform porous layer (case v

In Situ Hydrodynamic Spectroscopy of Spray-Pyrolyzed LiMn2O4 (LMO) Electrode of Complex Morphology

This study includes almost all the essential features of the recently developed EQCM-D methodology to monitor intercalation-induced dimensional and porous structure changes in battery electrodes.17 Using spray pyrolysis, we have fabricated LMO electrodes of different thicknesses and morphologies, as schematically shown at the top of Figure 6a. When characterized in air, Δf/n shifts due to the deposited electrode masses appearing to be n-independent with relatively small changes in ΔW/n, consistent with the model of a rigid electrode’s active mass. The difference in the electrode morphology is not seen by EQCM-D measurements in air because air has too low a viscosity. As expected, these thin electrodes, when immersed in Li2SO4 solution under OCV, 73

DOI: 10.1021/acs.accounts.7b00477 Acc. Chem. Res. 2018, 51, 69−79

Article

Accounts of Chemical Research

Figure 4. Electrodeposited Cu films of different porosities prepared by CV and CA (a and b, respectively). The related changes of Δf/n and ΔW/n (shown by different symbols) are plotted for the 3rd overtone versus the deposited charge (c and d, respectively). Straight lines (FFaraday) present frequency changes calculated using Faraday’s law and the Sauerbrey equation. Adapted with permission from ref 26. Copyright 2016, American Chemical Society.

Figure 5. (a) Different morphologies of Cu coatings obtained by CV and CA (top and middle of the panel) compared to that of Cu layer obtained by sputtering (bottom). (b) In situ hydrodynamic spectroscopy of these coatings: the circles correspond to different morphologies as indicated; solid lines are best fit curves describing hydrodynamic interactions for uniform porous layer and porous layer + bumps models, and dashed lines are the Kanazawa-type response of rigid layers with a flat surface. Adapted with permission from ref 26. Copyright 2016, American Chemical Society.

showed a small, gradually developing deviation of Δf/n and ΔW/n from the straight lines consistent with the presence of the relatively dense (nonporous) bottom layer covered by a thin top porous electrode layer (Figure 6b). It is also clearly

seen that the maximum deviation from the straight lines, especially for Δf/n, was observed for the thickest electrode in the series we prepared, whose top porous layer was additionally decorated with micrometer-sized bumps. In Figure 7c and d, 74

DOI: 10.1021/acs.accounts.7b00477 Acc. Chem. Res. 2018, 51, 69−79

Article

Accounts of Chemical Research

Figure 6. Sketches of LMO coatings of different thickness and morphology characterized by EQCM-D in air (a), Li2SO4 solution under OCV (b), and during Li-ion extraction/insertion for thinner and thicker coatings (c and d, respectively). Treatment of the responses for the different coatings by means of in situ hydrodynamic spectroscopy (e). Adapted with permission from ref 17. Copyright 2016, Nature Publishing Group.

Figure 7. SEM image of layered structure of an MXene flake (a). Long-term cycling of MXene electrode consisting of parallel arranged single layer particles under CV conditions in Li2SO4 solution (b). Suggested mechanism of viscoelastic changes in MXene electrode during insertion of water molecules and Li ions (c). Adapted with permission from ref 27. Copyright 2017, American Chemical Society.

the intercalation-induced Δf/n and ΔW/n shifts for the entire Li-ion extraction/insertion processes with the thinnest and thickest electrodes are presented. Because the thinnest (also called acoustically thin) electrode is composed of only the dense (nonporous) layer, the dissipation of the oscillation

energy is absent as the hydrodynamic interaction between the relatively flat surface of this electrode with the electrolyte solution phase is not changed during the Li-ion extraction/ insertion processes. As a consequence, n-independent frequency changes coincide with the frequency change 75

DOI: 10.1021/acs.accounts.7b00477 Acc. Chem. Res. 2018, 51, 69−79

Article

Accounts of Chemical Research

Figure 8. Changes in intercalation charge, frequency, and dissipation at the beginning and end of cycling (a) and in the middle of the cycling life during a complete Li insertion/extraction scan (b). Experimental changes of Δf/n and ΔD with overtone order both under OCV (deintercalated state) and in the fully intercalated state of the electrode were fitted by the viscoelastic model (c). Adapted with permission from ref 27. Copyright 2017, American Chemical Society.

Figure 9. Schematics of preparation of composite LFP electrode coatings containing large and small intercalation particles (a). Small and large particles are well-embedded or only partially embedded into the continuous binder network, respectively, revealing nonslip and slip conditions during crystal oscillations. Unprocessed experimental changes in Δf/n and ΔD as functions of time for different overtone orders after immersion of neat PVdF- and NaCMC-coated crystals into Li2SO4 solution (b and c, respectively). Adapted with permission from ref 28. Copyright 2017, American Chemical Society.

In contrast, the same Li-ion extraction/insertion process with the thickest electrode of the complex morphology results in large volume changes revealed by the strong n-dependence of the Δf/n and ΔW/n shifts (Figure 6d). A quantitative hydrodynamic analysis was performed using the difference

calculated from the intercalation/deintercalation charge using the Faraday law and the Sauerbrey equation. The resonance width does not change so that the entire process of Li-ion extraction/insertion into LMO is rigorously validated as an ideal gravimetric case (Figure 6c). 76

DOI: 10.1021/acs.accounts.7b00477 Acc. Chem. Res. 2018, 51, 69−79

Article

Accounts of Chemical Research

Figure 10. Characterization of composite LTO electrodes in ethylene carbonate + diethyl carbonate (EC+DMC) solutions of Li bis(trifluoromethanesulfonyl)imide (LiTFSI) and Li hexafluorophosphate (LiPF6): CVs for the first few cycles (a and b, respectively), SEM images of pristine and the related SEI (c−e, respectively), and diagram graphs of the calculated viscoelastic parameters (f, g, respectively). Adapted from ref 29.

spectra: first, the experimental Δf/n and ΔW/n shifts in the deintercalated state of LMO were presented taking the intercalated state as a reference, and second, the resulting experimental points were fitted to the related hydrodynamic models (symbols and dashed lines in Figure 6e, respectively). A reasonably good fit was achieved, and the small change of the intercalation-induced structural parameters was quantified (for details, see the original paper17).

Li2SO 4 solution, and the electrodes are subjected to continuous cycling (in both cases, the reference state relates to the neat crystal measured in air, see background section). The related viscoelastic parameters are listed and discussed in the original paper.27 The analysis of the changes in the viscoelastic parameters of MXene electrodes after their contact with water and during insertion of Li ions converge to the scheme shown in Figure 8c: insertion/extraction of water after water impregnation and complete drying of the electrodes, respectively, result in reversible softening/stiffening of the electrodes caused by specific hydrogen bonding in the MXene electrode interspaces decorated by =O and −OH functional groups. In contrast, insertion/extraction of Li ions into the electrode bulk results in their reversible stiffening/softening due to modification of the specific H-bonding of water by Li cations and contraction of the interspaces width caused by a large charge-to-size density. The highly reversible character of viscoelastic changes of MXene electrodes caused by insertion of Li-ions correlates well with the excellent capacity retention of these electrodes.27

Viscoelastic Changes in Ti3C2(OH)2 (MXene) Electrode

Figure 7a shows that MXene is characterized by a perfectly capacitive CV response that does not change after 100 cycles (Figure 7b).27 Figure 8a summarizes long-term monitoring of the changes in charge, Δf/n and ΔD, for different overtone orders as functions of the cycle number (from top to bottom). Figure 8b presents an enlarged view of the same quantities as a function of time for a complete Li-ion insertion/extraction cycle. Strong n-dependent changes create a certain pattern (signature) of the Δf/n and ΔD plots as functions of n. The experimental values of Δf/n and ΔD and the results of the viscoelastic modeling are shown in Figure 8c for two selected electrode states: (i) when MXene electrodes are transferred from air into pure water and (ii) when water is replaced by 77

DOI: 10.1021/acs.accounts.7b00477 Acc. Chem. Res. 2018, 51, 69−79

Article

Accounts of Chemical Research Intercalation-Induced Viscoelastic Changes in the Composite LiFePO4 (LFP) Electrode Containing Stiff and Soft Binders

viscoelastic properties of SEI on Sn anode in the presence of different electrolyte solutions has been pioneered by a research group at Argonne National Laboratory.10



In a recently published paper, we have described continuous monitoring of the viscoelastic changes in composite LFP electrodes containing intercalation particles of different sizes that are nonuniformly embedded into the 3D network of the polymeric binder (Figure 9a).28 The stiff character of PVdF and soft nature of NaCMC neat films in aqueous solution of Li2SO4 are already seen from the raw Δf/n and ΔD data (ndependent changes in Δf/n and ΔD unchanged with time, and the characteristic n-dependent changes of Δf/n and ΔD for PVdF and NaCMC, respectively (Figure 9b and c, respectively). For details about EQCM-D signatures of the composite LFP electrode with stiff PVdF and soft NaCMC binders, and their peculiar dependence on the LFP particle size, see the original recent paper.28

CONCLUSIONS AND PERSPECTIVES We have developed a novel in situ real-time EQCM-D-based methodology to monitor gravimetric, dimensional, and porous structure changes in stiff electrodes and viscoelastic changes in soft electrodes. Fast screening of electrolyte solutions components for the formation of high-quality SEI on LTO electrodes has been demonstrated as one of the successful applications of EQCM-D to the battery field. In the near future, we expect this method to be developed in three major directions specifically for the needs of energy-storage science and technology. (i) The method is well-adjusted to distinguish between the nanoconfined and moveable fluid in supercapacitor electrodes such as 2D-layered materials with fast ionic transport. The combination of EQCM-D with ac electrogravimetry for this purpose would be the most effective. (ii) Understanding of the origin of low-frequency/static intercalation-induced stresses in alloy-type electrodes with high volume change using multiharmonic EQCM-D, especially in combination with MOSS (multibeam optical stress sensor method) is required. This may help to throw new light on the mechanical behavior and failure mechanisms in cycled highspecific capacity Si anodes. (iii) Development of EQCM-D theory for interpreting viscoelastic behavior of nonconventional electrode systems such as particulate composite electrodes, nanostructured and foam electrodes.

Quantification of SEI Formed on the Surface of the LTO Anode

Protective solid-electrolyte interfaces (SEI) are formed on most Li battery electrodes playing crucially important roles for their reversible and stable operation.1 Good SEI films should be thin and dense, ensuring high mechanical integrity, and high Li-ion conductivity but possessing negligibly small electronic conductivity. As an illustrative case, we choose the Li4Ti5O12 (LTO) anode, which is considered a “zero-stress” material. Hence, the measured EQCM-D parameters can be attributed to the (gravimetric) intercalation-induced changes in the electrode host and to the combined gravimetric and viscoelastic changes in the generated surface films.29 For understanding how the electrode stability is influenced by the mechanical nature of the SEI, LTO electrodes were examined in three commonly used electrolyte solutions: LiTFSI, LiPF6, and LiPF6 + 2% VC in EC-DMC solvent mixtures. CVs of LTO electrodes in LiTFSI and LiPF6 solutions are shown in Figure 10a and b. HR SEM images indicative of SEI formation are shown in Figure 10c−e (raw EQCM-D are presented in the original paper29). Quantification of the mechanical parameters of the resulting SEI using viscoelastic modeling is indicative of a much higher quality of SEI formed on LTO anodes in LiTFSI solutions expressed by the highest storage modulus and the lowest thickness compared to the situations in the other electrolyte solutions (Figure 10f, g). The mechanical properties of the SEI formed in different electrolyte solutions were first tested in semiflooded EQCM-D cells and then reconfirmed in similar experiments in coin cells under starved-electrolyte solution conditions. Much better capacity retention in the LTO/LiTFSI system, and gradual capacity deterioration of the LTO/LiPF6 system linked to a poorer SEI quality, were observed in both types of experiments. However, in the semiflooded EQCM cells, selection of the better electrolyte solution for the LTO cycling can be made after a few first fast cycles (15−20 min total duration time), whereas in coin cells with starved electrolyte solution tests, the difference in the electrolyte solutions is seen only after 100 slow-rate cycles (200 h total duration time). These findings demonstrate the benefits of EQCM-D as a methodology for fast screening of the solution components, particularly for prediction of the SEI quality in the presence of different electrolyte solutions and additives during short-time diagnostics. A similar approach to analysis of



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Doron Aurbach: 0000-0001-8047-9020 Funding

This work has been supported by the Israel Ministry of Science Technology and Space Grant 66032. N.S. thanks the Israel Ministry of Science Technology and Space for their financial support. Notes

The authors declare no competing financial interest. Biographies Netanel Shpigel is a Ph.D. student at Bar-Ilan University guided by Profs. M.D. Levi and D. Aurbach. His main topic is the advanced application of EQCM-D and AFM for characterization of energy storage materials. Mikhael D. Levi is a Professor at Bar-Ilan University working in the lab headed by Prof. D. Aurbach. He received his Ph.D. in 1976 under the supervision of Academician Professor A.N. Frumkin in Moscow State University. He specializes in the development of fine electroanalytical methods for characterizations of battery materials. Since 2009, he has been developing a nongravimetric EQCM method adjusted for simultaneous tracking of gravimetric, viscoelastic, and porous structure changes in battery materials. 78

DOI: 10.1021/acs.accounts.7b00477 Acc. Chem. Res. 2018, 51, 69−79

Article

Accounts of Chemical Research

(16) Daikhin, L.; Sigalov, S.; Levi, M. D.; Salitra, G.; Aurbach, D. Quartz Crystal Impedance Response of Nonhomogenous Composite Electrodes in Contact with Liquids. Anal. Chem. 2011, 83, 9614− 9621. (17) Shpigel, N.; Levi, M. D.; Sigalov, S.; Girshevitz, O.; Aurbach, D.; Daikhin, L.; Pikma, P.; Marandi, M.; Jänes, A.; Lust, E.; Jäckel, N.; Presser, V. In Situ Hydrodynamic Spectroscopy for Structure Characterization of Porous Energy Storage Electrodes. Nat. Mater. 2016, 15, 570−575. (18) Urbakh, M.; Daikhin, L. Surface Morphology and the Quartz Crystal Microbalance Response in Liquids. Colloids Surf., A 1998, 134, 75−84. (19) Daikhin, L.; Urbakh, M. Effect of Surface Film Structure on the Quartz Crystal Microbalance Response in Liquids. Langmuir 1996, 12, 6354−6360. (20) Etchenique, R.; Calvo, E. J. Electrochemical Quartz Crystal Microbalance Gravimetric and Viscoelastic Studies of Nickel Hydroxide Battery Electrodes. J. Electrochem. Soc. 2001, 148, A361− A367. (21) Voinova, M. V.; Rodahl, M.; Jonson, M.; Kasemo, B. Viscoelastic Acoustic Response of Layered Polymer Films at FluidSolid Interfaces: Continuum Mechanics Approach. Phys. Scr. 1999, 59, 391−396. (22) Escobar-Teran, F.; Arnau, A.; Garcia, J. V.; Jiménez, Y.; Perrot, H.; Sel, O. Gravimetric and Dynamic Deconvolution of Global EQCM Response of Carbon Nanotube Based Electrodes by AcElectrogravimetry. Electrochem. Commun. 2016, 70, 73−77. (23) Hillman, A. R.; Efimov, I.; Skompska, M.; Wudl, F.; Bandey, H. L. Dynamics of Regioregular Conducting Polymer Electrodes in Response to Electrochemical Stimuli. Faraday Discuss. 2002, 121, 423−439. (24) Ispas, A.; Bund, A. Electrochemical Quartz Crystal Microbalance. Encylcopedia of Applied Electrochemistry 2014, 554. (25) Taviot-Guého, C.; Vialat, P.; Leroux, F.; Razzaghi, F.; Perrot, H.; Sel, O.; Jensen, N. D.; Nielsen, U. G.; Peulon, S.; Elkaim, E.; Mousty, C. Dynamic Characterization of Inter- and Intralamellar Domains of Cobalt-Based Layered Double Hydroxides upon Electrochemical Oxidation. Chem. Mater. 2016, 28, 7793−7806. (26) Sigalov, S.; Shpigel, N.; Levi, M. D.; Feldberg, M.; Daikhin, L.; Aurbach, D. Electrochemical Quartz Crystal Microbalance with Dissipation Real-Time Hydrodynamic Spectroscopy of Porous Solids in Contact with Liquids. Anal. Chem. 2016, 88, 10151−10157. (27) Shpigel, N.; Lukatskaya, M. R.; Sigalov, S.; Ren, C. E.; Nayak, P.; Levi, M. D.; Daikhin, L.; Aurbach, D.; Gogotsi, Y. In Situ Monitoring of Gravimetric and Viscoelastic Changes in 2D Intercalation Electrodes. ACS Energy Lett. 2017, 2, 1407−1415. (28) Dargel, V.; Jäckel, N.; Shpigel, N.; Sigalov, S.; Levi, M. D.; Daikhin, L.; Presser, V.; Aurbach, D. In Situ Multilength-Scale Tracking of Dimensional and Viscoelastic Changes in Composite Battery Electrodes. ACS Appl. Mater. Interfaces 2017, 9, 27664− 27675. (29) Dargel, V.; Shpigel, N.; Sigalov, S.; Nayak, P.; Levi, M. D.; Daikhin, L.; Aurbach, D. In Situ Real-Time Gravimetric and Viscoelastic Probing of Surface Films Formation on Lithium Batteries Electrodes. Nat. Commun. 2017, 8, 1389.

Sergey Sigalov is a postdoctoral student at Bar-Ilan University guided by Profs. M.D. Levi and D. Aurbach. He received his Ph.D. at Bar-Ilan University and continues work on the use of EQCM-D for characterization of battery and supercapacitor electrodes. Leonid Daikhin is a Professor at Tel Aviv University. He is involved in theoretical work related to hydrodynamic and viscoelastic modeling. Doron Aurbach is a full professor, senate member, and head of the electrochemistry group at Bar-Ilan University. He leads the Israel National Research Center for Electrochemical Propulsion (INREP; 22 research groups) and is an MRS, ECS, and ISE fellow. He is involved in all aspects of Li, Na, Mg, Li−S, and Li−O2 battery research, the development of new electrolyte solutions, surface and materials science, new analytical methodologies, and supercapacitors.



REFERENCES

(1) Goodenough, J. B.; Park, K. S. The Li-Ion Rechargeable Battery: A Perspective. J. Am. Chem. Soc. 2013, 135, 1167−1176. (2) Arnau, A.; Antonio, A. Review of Interface Electronic Systems for AT-Cut Quartz Crystal Microbalance Applications in Liquids. Sensors 2008, 8, 370−411. (3) Bandey, H. L.; Martin, S. J.; Cernosek, R. W.; Hillman, A. R. Modeling the Responses of Thickness-Shear Mode Resonators under Various Loading Conditions. Anal. Chem. 1999, 71, 2205−2214. (4) Daikhin, L.; Gileadi, E.; Katz, G.; Tsionsky, V.; Urbakh, M.; Zagidulin, D. Influence of Roughness on the Admittance of the Quartz Crystal Microbalance Immersed in Liquids. Anal. Chem. 2002, 74, 554−561. (5) Efimov, I.; Gruia, V.-T.; Rumiche, F.; Bund, A.; Ispas, A. An Electrochemical Quartz Crystal Microbalance Study on Adsorption of Single Walled Carbon Nanotubes onto poly[3,4-Ethylenedioxythiophene] Layers. J. Solid State Electrochem. 2015, 19, 2581−2589. (6) Johannsmann, D. The Quartz Crystal Microbalance in Soft Matter Research; Springer, 2015. (7) Rodahl, M.; Höök, F.; Fredriksson, C.; Keller, C. A.; Krozer, A.; Brzezinski, P.; Voinova, M.; Kasemo, B. Simultaneous Frequency and Dissipation Factor QCM Measurements of Biomolecular Adsorption and Cell Adhesion. Faraday Discuss. 1997, No. 107, 229−246. (8) Hillman, A. R.; Mohamoud, M. A.; Efimov, I. Time− Temperature Superposition and the Controlling Role of Solvation in the Viscoelastic Properties of Polyaniline Thin Films. Anal. Chem. 2011, 83, 5696−5707. (9) Yang, Z.; Dixon, M. C.; Erck, R. A.; Trahey, L. Quantification of the Mass and Viscoelasticity of Interfacial Films on Tin Anodes Using EQCM-D. ACS Appl. Mater. Interfaces 2015, 7, 26585−26594. (10) Sauerbrey, G. Verwendung von Schwingquarzen Zur Wägung Dünner Schichten Und Zur Mikrowägung. Eur. Phys. J. A 1959, 155, 206−222. (11) Buttry, D. a.; Ward, M. D. Measurement of Interfacial Processes at Electrode Surfaces with the Electrochemical Quartz Crystal Microbalance. Chem. Rev. 1992, 92, 1355−1379. (12) Marx, K. a. Quartz Crystal Microbalance: A Useful Tool for Studying Thin Polymer Films and Complex Biomolecular Systems at the Solution-Surface Interface. Biomacromolecules 2003, 4, 1099− 1120. (13) Tsai, W.-Y.; Taberna, P.-L.; Simon, P. Electrochemical Quartz Crystal Microbalance (EQCM) Study of Ion Dynamics in Nanoporous Carbons. J. Am. Chem. Soc. 2014, 136, 8722−8728. (14) Keiji Kanazawa, K.; Gordon, J. G. The Oscillation Frequency of a Quartz Resonator in Contact with Liquid. Anal. Chim. Acta 1985, 175, 99−105. (15) Levi, M. D.; Shpigel, N.; Sigalov, S.; Dargel, V.; Daikhin, L.; Aurbach, D. In Situ Porous Structure Characterization of Electrodes for Energy Storage and Conversion by EQCM-D: A Review. Electrochim. Acta 2017, 232, 271−284. 79

DOI: 10.1021/acs.accounts.7b00477 Acc. Chem. Res. 2018, 51, 69−79