Correlative Electrochemical Microscopy of Li‐Ion (De)intercalation at a Series of Individual LiMn2O4 Particles

Abstract The redox activity (Li‐ion intercalation/deintercalation) of a series of individual LiMn2O4 particles of known geometry and (nano)structure, within an array, is determined using a correlative electrochemical microscopy strategy. Cyclic voltammetry (current–voltage curve, I–E) and galvanostatic charge/discharge (voltage–time curve, E–t) are applied at the single particle level, using scanning electrochemical cell microscopy (SECCM), together with co‐location scanning electron microscopy that enables the corresponding particle size, morphology, crystallinity, and other factors to be visualized. This study identifies a wide spectrum of activity of nominally similar particles and highlights how subtle changes in particle form can greatly impact electrochemical properties. SECCM is well‐suited for assessing single particles and constitutes a combinatorial method that will enable the rational design and optimization of battery electrode materials.

Asapromising Li-ion battery cathode material in both aqueous and organic electrolytes,s pinel LiMn 2 O 4 has attracted much attention in recent years owing to its large theoretical capacity,h igh abundance,a nd nontoxicity, [1] although an umber of problems remain to be resolved. [2] As with much research in electrochemistry,m acroscale electrochemical measurements have mainly been used to study battery materials,w hich for complex composite electrodes include contributions from the conductive agent, adhesive, and the active material. [3] Local structure-activity relationships for battery materials are still somewhat unexplored in regards to individual active particles or particle agglomerates. [4] Indeed, the variation in activity among particles,a nd the relation to particle topology and structure,h as largely remained elusive.T his paper addresses this issue head on, through the use of as trategy that enables the measurement and direct comparison of the structure and electrochemical activity of individual particles.
In order to rationally design battery electrodes,a nd electroactive materials in general, strategies that enable the direct correlation between the local redox activity and electrode structure are highly valuable. [5] In addition to some optical approaches (e.g., plasmonic imaging [6] ), emerging in situ scanning electrochemical probe microscopy (SEPM) [7] techniques also promise to provide insight into the structural factors controlling the electrochemical behavior of battery electrode materials.W ithin the SEPM family, scanning electrochemical microscopy (SECM) has been the most widely used in Li-ion battery research, especially for probing the electrically insulating solid electrolyte interface (SEI), although mainly on the scale of tens of microns. [8] Scanning ion conductance microscopy (SICM) offers amuch higher spatial resolution and has been used to visualize ionflux spatial heterogeneities in tin and silicon anodes in Li-ion batteries. [9] It is worth noting that both SECM and SICM collect electrochemical information about an electrode substrate by monitoring the spatially dependent concentrations/ fluxes of the reactant, product, or intermediates at ascanning electrode tip.
In contrast, in scanning electrochemical cell microscopy (SECCM), electrochemistry is probed directly and locally at asubstrate electrode,with aspatial resolution defined by the area of meniscus contact, and with the possibility of synchronous co-location topographical mapping. [5a, 10] In the context of battery research, this technique has previously been used to electrochemically interrogate thin films of (insulating) Li 2 O 2 , [11] as well as small populations of LiFePO 4 particles (ca. 10 particles). [4,12] In this study,SECCM has been deployed in as ingle-channel nanopipette configuration to investigate the electrochemical behaviour of individual LiMn 2 O 4 particles within an ensemble,w hich were visualized, post-experiment, by co-located SEM. Experimental details are available in the Supporting Information, Section S1. SECCM was deployed in hopping mode, [5a,13] as shown schematically in Figure 1a (labelled in the Supporting Information, Section S2, Figure S1). In this configuration, ananopipette probe,containing 1.0 m aqueous LiCl as the electrolyte and aA gCl-coated Ag wire as aq uasi-reference counter electrode (QRCE), was approached to the substrate (working electrode) surface to make meniscus contact at as eries of predefined locations in ag rid ( Figure S2). At each landing, local electrochemical measurements (I-E or E-t)were made within the confined area defined by the meniscus cell (the probe itself did not make physical contact with the surface). Thes ubstrate was prepared by drop casting spinel LiMn 2 O 4 particles onto glassy carbon (GC;F igure S3).
To explore the Li + storage mechanism at individual LiMn 2 O 4 particles,aswell as visualize the variation in activity within an active ensemble,s patially resolved cyclic voltammetry was performed on the as-prepared LiMn 2 O 4 /GC electrode.S tarting at 0V vs.A g/AgCl, the potential was swept between 0t o1 .25 Vatar ate of 1Vs À1 .A ss hown in Figure 1b,arelatively featureless cyclic voltammogram (CV) was obtained on the GC support, with processes encountered at extreme anodic and cathodic potentials attributable to carbon corrosion [14] and the oxygen reduction reaction (ORR), [15] respectively.T hus,t he electrochemical stability window of GC was estimated to be approximately 1.8 Vunder these conditions. Figure 1c depicts arepresentative CV obtained at asingle LiMn 2 O 4 particle,e ncapsulated by the meniscus (droplet) cell. Li + (de)intercalation chemistry at LiMn 2 O 4 can be expressed by Eq. (1): where typically 0 < x < 1. [16] During the charging process [Eq.
(1), forward],Li + is extracted from the structural framework of LiMn 2 O 4 ,c oinciding with the oxidation of Mn III to Mn IV .T his corresponds to the sweep in the positive direction, in which two redox peaks located at 0.89 and 1.01 Vv s. Ag/ AgCl (1.0 m LiCl) can be assigned to Li + extraction from tetrahedral lattice sites in the presence and absence of the Li-Li interaction, respectively. [16] Thereverse processes [Eq.
(1), reverse] occurred during discharge,with the two peaks at 0.69 and 0.89 Vi nt he negative sweep corresponding to the two different Li + insertion processes.I na ddition, no undesirable side (parasitic) reactions were observed at high potentials, demonstrating that the oxygen evolution reaction does not occur on LiMn 2 O 4 in this potential range. [17] It is interesting to note that this scan rate (n)i s2 -4 orders of magnitude larger than that employed in bulk electrochemical experiments with the same material (n = 0.1-10 mV s À1 ), [18] and yet the (de)intercalation processes are facile.T his indicates that in the traditional composite electrode configuration, the achievable (de)intercalation rates are largely governed by the rate of electron transfer between the auxiliary elements (e.g., binder and carbon black) and electroactive components (see below). Note that the low currents passed during measurement in SECCM makes it relatively immune to resistance arising from the sample itself (e.g., low intrinsic conductivity or contact resistance), making this technique ideal for the study of adiverse range of (semi)conductive materials. [5a,19] Individual LiMn 2 O 4 particles within the ensemble were probed in an automated fashion by performing ah opping mode SECCM scan in the voltammetric mode,inwhich each hop corresponds to an independent, spatially resolved CV experiment. [5a, 13] Thehopping distance (i.e., distance between each landing/pixel) was 1.5 mm, which ensured each measurement spot was independent of the last. An SEM image of the probed area, post-scan, is shown in Figure 2a (also shown enlarged in Figure S4). Evidently,the probed area is predominantly GC (individual droplet footprints are visible in the scan area), with ac ollection of LiMn 2 O 4 particles scattered throughout. Ac omparison with the SECCM topographical (z-height) map in Figure 2b,revealed 18 pixels with elevated topography,e ach corresponding to an isolated LiMn 2 O 4 particle or agglomerate (see below). Thec o-location of the particles (Figure 2a)and the higher points in the topography map ( Figure 2b)g ives us confidence that the SECCM technique can be used to identify particles in situ.
Aspatially resolved CV-SECCM movie (current maps as af unction of potential) obtained on the LiMn 2 O 4 /GC ensemble electrode (60 60 mm 2 ,4 0 40 pixels) is shown in the Supporting Information, Section S3 and Movie S1. The magnitude of the anodic and cathodic currents (i.e., peak current) obtained at each individual active pixel is comparable throughout, signifying that Li + (de)intercalation is relatively reversible (see below). Figure 2c,d depicts two frames from the movie,taken from the anodic (forward) and cathodic On the right is an enlarged diagram of the probe-particle-support interface at asingle pixel of ascanning experiment, in which an individual LiMn 2 O 4 particle is fully encapsulated by the meniscus cell. b) Four SECCM CVsobtained at the GC support and c) atypical CV obtained from asingle LiMn 2 O 4 particle. Experiments performed in 1 m LiCl, with a500 nm diameter probe, at ascan rate (n)of1Vs À1 . (reverse) sweeps at potentials of 1.0 Vand 0.6 V, respectively. Through correlation of the activity maps with the SEM image of the scan area (Figure 2a)a nd surface topography map (Figure 2b), it is obvious that the individual LiMn 2 O 4 particles exhibit elevated currents compared to the relatively inert GC support. It should be noted that while aC V-scan hopping protocol was employed above,chronoamperometric (current-time curve, I-t)w aveforms can also be applied if only as ingle potential is of interest, as shown in Figure S5.
Thei ndividual LiMn 2 O 4 particles (including primary particles and agglomerated secondary particles) exhibit very different current magnitudes in Figure 2c,d, indicative of heterogeneous size and activity within the ensemble.Indeed, by extracting the individual CVs from each active pixel, as shown in Figure S6, it is clear that each particle/agglomerate presents au nique I-E profile,a ttributable to its physical heterogeneities (e.g., particle size,c omposition, crystallinity, or orientation), as demonstrated by the corresponding highresolution SEM images in Figure S7. It is worth reemphasizing, the variation in I-E characteristics ("activity") among superficially similar particles (or agglomerates) is completely invisible in macroscopic (bulk) measurements,w hich reflect the average response of the ensemble (see below). As the probed area (indicated by the individual droplet footprints) is only al ittle bit larger than the tip diameter (500 nm, Figure S8), some LiMn 2 O 4 agglomerates cannot be fully encapsulated by the SECCM meniscus.I no rder to treat the data semi-quantitatively (i.e., the active particle surface area is known, see below), the meniscus cell should totally encapsulate the particle during measurement, as shown schematically in Figure 1a.T hus,m ultiple scans were performed on different areas of the LiMn 2 O 4 /GC ensemble and only pixels in which particles were small (or sparse) enough to be fully encapsulated by the meniscus were selected for comparison and quantitative analysis,asdepicted in Figure 3. Afurther indication of the validity of this approach is that the overall peak currents fall within afairly narrow range of circa 30-70 pA, notwithstanding some variation in the peak potentials and overall CV morphology.N ote that the size of the nanopipette probe could easily be tailored to accommodate encapsulation of larger particles,o rs maller particle-toparticle separations.
Them agnitude of the current measured at each pixel is governed by the size (i.e., the exposed surface area) of the Figure 3. a-h) CVs( i) and corresponding SEM images (ii)f rom individualL iMn 2 O 4 particles supported on GC. The CV measurements (n = 1Vs À1 ) were obtained by local ensemble measurements with SECCM, with a500 nm diameter probe filled with 1 m LiCl. LiMn 2 O 4 particle,while the position and shape of the anodic and cathodic peaks,i ndicative of the Li + (de)intercalation mechanism and kinetics,i sg overned by the particular properties (i.e., composition, crystallinity,a nd orientation) of the particle.Itisimportant to note that the electrochemical behaviour of individual LiMn 2 O 4 particles is highly heterogeneous,w ith the voltammetric peak morphology (position, separation, and width) varying considerably throughout the ensemble.Some particles,such as particles (g) and (h), exhibit very sluggish kinetics (i.e., large peak-to-peak separations), which is not desirable for the application of this material as an active battery material. By comparison, particle (c), which appears to be comprised of small crystallographic facets, exhibits fast kinetics,making it the ideal structure that should be pursued through the application of novel design principles. To further illustrate this point, detailed comparisons of the electrochemical properties (voltammetric peak potential and current, total charge and cathodic-to-anodic charge ratio) of each individual particle in Figure 3a re summarized in Table S1. Ap articularly interesting observation is that the cathodic-to-anodic charge ratio (calculated by dividing the total cathodic charge by the total anodic charge) is higher than 100 %f or all particles,w hich is ascribed to the Jahn-Te ller effect. [20] In brief,afraction of the Mn 3+ is further reduced to Mn 2+ during the reverse scan (Li + intercalation process), which subsequently undergoes dissolution into the electrolyte.T hus,t he material is over-reduced, resulting in enhanced cathodic charge and an apparent cathodic-toanodic charge ratio greater than 100 %d uring cycling.A s the CV measurement only probes the near-surface processes (i.e., only 10-30 %o ft he total capacity can be used), this phenomenon can carry on for multiple cycles without apparent capacity loss ( Figure S9). Besides this,t he voltammetric peak-to-peak separation (DE p )isobserved to decrease during the multiple voltammetric cycling, indicating that the (de)intercalation processes become kinetically more facile at the single particle level.
To further clarify the relationship between single particle and conventional macroscale electrochemistry,v oltammetry was performed on ac omposite (i.e., material, binder,a nd conductive additive) LiMn 2 O 4 electrode ( Figure S10). Note that in bulk only af raction of the total capacity is accessed (e.g.,2 3% at 5mVs À1 )a nd the cathodic-to-anodic charge ratio is greater than 100 %, in agreement with the singleparticle measurements above,a swell as previous reports. [21] Viewing these results alongside those from SECCM (Figure 3), it is very clear that the bulk electrochemical response "washes out" the unique properties of each individual LiMn 2 O 4 particle.T his contrasts with the SECCM measurements,w hich reveals the heterogeneity of activity at the single particle level. To illustrate this point further,t he 8 CVs in Figure 3w ere averaged ( Figure S11) to produce ac urve that superficially resembles (i.e., two anodic peaks observed at 0.8 and 1.0 V) the bulk ensemble response.I t should also be noted that the bulk composite electrode response can be reproduced in the SECCM configuration at low n using large,m icrometric probes (8 and 50 mmi n diameter, Figure S8), in which ac ollection of LiMn 2 O 4 particles are probed during each experiment ( Figure S12). This demonstrates that the diversity of responses observed in Figure 3m ust arise from intrinsic differences between the LiMn 2 O 4 particles,r ather than being an artefact of the SECCM configuration or the high n used, again underscoring the importance of kinetic effects in Li + (de)intercalation reactions.
To complete this study and highlight further the versatility of the SECCM approach, spatially-resolved galvanostatic charge-discharge measurements were performed at the single particle level, with an applied current of AE 5pAfor 1sat each measurement point. Spatially resolved, potential-time snapshots (maps) obtained at different times and current polarities are presented in Figure S13 a-d. Again, by comparing the maps with the corresponding SEM image in Figure S13 e, it is clear that different particles present different charge/discharge potentials,a ttributed to unique structural characteristics (i.e., size and morphology). Figure S13 fshows arepresentative E-t curve (galvanostatic charge/discharge profile) extracted from asingle LiMn 2 O 4 particle,inwhich the charge/ discharge processes occur at apotential of circa 0.75 Vvs. Ag/ AgCl, which is consistent with the peak position in the CVs shown in Figure 3. In contrast, at GC,the measured potential changes rapidly (non-faradaic or capacitive charging current) before reaching the electrochemical window limits highlighted in Figure 1b,a se xpected for an ideal polarizable electrode system. Figure 4depicts the galvanostatic charge-discharge measurements performed on individual LiMn 2 O 4 particles (agglomerates) that again, are small enough to be fully encapsulated by the SECCM meniscus (electrochemical cell). In line with the CV results above,e ach particle presents au nique E-t profile,w ith different charge/discharge potentials and ohmic (IR,w here R is resistance) drops (i.e., the potential difference between the charge/discharge plateau), as summarized in Table 1. Again, it needs to be reiterated that the heterogeneity in activity (E-t profiles in Figure 4o rC Vs in Figure 3) among superficially similar LiMn 2 O 4 particles or agglomerates is al argely unexplored phenomenon that is obscured in traditional macroscopic measurements on composite electrodes.I ts hould also be noted that the IR drop values are very low,especially considering the extremely high charge/discharge rates implemented in this study (e.g., the IR drop was only ca. 20 mV at aC-rate of 279 Cfor particle bin Figure 4). This value is among the highest C-rates reported in the literature,with high rate performance Zn (up to 50 C) and Al (up to 500 C) ion battery electrodes being reported before. [22] As alluded to above,t his indicates that in the traditional composite electrode configuration, IR drop (and hence rate-performance limitation) is largely governed by the rate of electron transfer between the auxiliary elements (e.g., binder and carbon black) and electroactive component(s), rather than Li + (de)intercalation into the individual LiMn 2 O 4 particles.T hus,t here remains great potential to further improve the rate capability in battery electrochemistry by new strategies to wire active particles or by improving the electrode preparation method to enhance the charge transfer kinetics (see above). [23] It needs to be reiterated that the timescale of these localized E-t experiments is orders-ofmagnitude faster than that usually encountered in bulk electrochemical measurements (i.e., 0.1 to 10 Cr ates), which is explored in detail in Section S4.
In summary,u sing amobile meniscus cell in the SECCM configuration, we have been able to probe and compare the electrochemical activities of individual particles within an ensemble.T his direct and local probe method has enabled characteristic features to be targeted and analysed precisely through acorrelative approach with ex situ SEM. Specifically, in this work LiMn 2 O 4 ,apromising Li-ion battery cathode material, has been revealed to possess significantly heterogeneous electrochemical behaviour [i.e., Li + (de)intercalation processes] at the single particle level, attributable to differences between particle size,c omposition, crystallinity,a nd orientation. In addition, the variation in electrochemical activity revealed by these sub-microscale (single particle) measurements has allowed us to rationalize the macroscopic bulk electrochemical response of complex composite battery electrodes.
In the past few years,anumber of in situ/in operando analysis tools have been established for the exploration of complex redox processes in battery materials. [24] However,to date,there have been relatively few reports of techniques that can provide information at single particle level or possess the capability to distinguish variations in the electrochemical performance of individual active entities.The work presented herein demonstrates new capabilities of SECCM, which pave the way for the deep investigation of electrode reaction processes in energy conversion/storage technologies.I nt he future,w ea im to visualize any minute influence of (nano)structure (e.g.,crystallographic orientation) on redox activity and (de)intercalation kinetics through ac ombination of rational materials design and synthesis [25] and SECCM. This will be achieved by investigating mono-dispersed particles on TEM grids and then performing characterization by highresolution analytical TEM.  [a] Particle labels correspond to those in Figure 4. [b] The volume of each particle was estimated based on the height (estimatedf rom z-height topography),w idth, and length (estimatedf rom SEM image) by assuming the particle is an ellipsoid (V = 4/3pabc).
[c] The capacity calculationprocess can be found in the Supporting Information,Section S4.