Shan
Yang
,
Binggong
Yan
,
Li
Lu
and
Kaiyang
Zeng
*
Department of Mechanical Engineering, National University of Singapore, Engineering Drive 1, Singapore 117576. E-mail: mpezk@nus.edu.sg; Fax: +65 6779 1459; Tel: +65 6516 6627
First published on 27th September 2016
This paper presents the results of in situ characterization of grain boundary effects on Li-ion diffusion in Li1.2Co0.13Ni0.13Mn0.54O2 thin film cathode by using various Scanning Probe Microscopy (SPM) techniques. In particular, conductive-AFM results show that grain boundaries are more conductive than those in the grain interior. With the increase of bias voltage, the high conductive regimes extend from grain boundaries to interiors. I–V curves show decreased current and increased voltage for current initiation when the tip is moved farther away from boundaries. Furthermore, positive and negative bias applied at grain boundary by biased-AFM can distinguish and manipulate the local Li-ion intercalation/de-intercalation processes at grain level in the cathode material without assembly of a full battery cell. Exfoliation and delamination, degradation and structural changes are observed when the Li-ions are move-out or move-into the layered structure of the cathode at the grain level. These results can provide important insights into understanding the Li-ion diffusion and aging mechanisms of cathode materials during charge/discharge processes.
Conductive atomic force microscopy (c-AFM) is a current-based technique for characterizing the conductance or resistance variations by detecting the current passing through the tip and the sample. In the c-AFM module, the specially designed cantilever holder consists of a transimpedance amplifier. During the measurement, a constant voltage is applied between a conductive tip and an electrode at the back of the sample. The current at the tip contact position is enhanced by the amplifiers, enabling the detection of small current. The standard c-AFM module is capable of measuring currents from ∼0.5 pA to 10 nA, allowing the conductivity measurement in the materials with relatively high resistivity.8 In addition, c-AFM can simultaneously map the surface topography and current, therefore the relationship between the topography and conductance can be obtained. This technique has been utilized in ion-conductive materials such as LiCoO2, LiMn2O4, LiNi0.8Co0.2O2 and Li-ion conducting glass ceramics.9–14 Furthermore, by positioning the AFM tip at a point of interest and applying a voltage ramp, the local current–voltage (I–V) curve at that point can be obtained.
Biased-AFM is another technique and it is used to apply DC bias to the sample surface through a conductive tip that is directly contacted with the sample. This technique is similar to the poling processes using the commercial piezoresponse force microscope with the primary difference in the bias-strain coupling mechanisms. During the measurement, the biased AFM tip concentrates an electric field in a nanometer-scale volume of material, inducing interfacial electrochemical processes at the tip–surface junction and ionic current flow in the material. The intrinsic link between ionic concentration, electrochemical reaction of the ions and the molar volume of the materials results in the surface deformation which is detectable by AFM probe.15 This technique has been previously used to study the effects of the cycling processes in all-solid-state thin film battery.16,17 In this work, biased-AFM measurements are performed with the positive and negative biases applied in a sequence at a single point with bias time of 60 seconds. This time duration is long enough to cause the Li-ions diffusion in the materials. The biased point and surrounding area are then scanned using the tapping mode AFM to detect the surface topography variations induced by the applied bias. Therefore, the processes of Li-ions diffuse into or out of the layered structure in cathode material can be studied separately, and these processes cannot be unambiguously distinguished by applying the AC voltage to the specimens.
During cycling process, Li-ions diffusion can result in conductance changes and molar volume changes in the electrode materials.18,19 Poor electrical conductivity of the cathode material declines its electrochemical performance and restricts its practical application. In addition, the irreversible topography changes can result in capacity fading, power decay and impedance increase (battery aging).20,21 It is therefore important to understand the mechanisms of Li-ions diffusion in order to develop new generation battery materials as well as to increase the reliability of the LIBs. It was expected that Li-ion diffusion occurs primarily through the grain boundaries in the cathodes materials with higher percentage of the grains having (003) orientation.22,23 First principles calculations have also shown that Li-ions diffusion along the grain boundaries as they have showed lower activation energy than that across the boundary,24 and the energy barrier of the boundary area was lower than that of the grain interior.10 However, the critical role of the grain boundary in Li-ions diffusion has not been well visualized due to lack of suitable characterization techniques. In this study, the grain boundary effects on Li-ions diffusion, in terms of conductance and surface morphology changes, are studied by using bias-AFM to first apply the bias, and then using tapping-mode AFM to observe the topography changes, also by using c-AFM to measure the conductance changes. This work aims to provide the fundamental understanding for constructing and demystifying the knowledge of the structure–property–functionality relationships in the LIBs cathode materials.
Among all of the cathode materials, the layered-structure cathodes have been extensively studied due to their capability of delivering high energy density and low cost. In recent years, layered oxide material Li2MnO3–LiMO2 (M = Ni, Co, Mn) is proposed to be a promising cathode system, as it have the capability to deliver high energy density (280 mA h g−1) which is approximately twice of that of the commercial cathode materials.25–27 In this work, Li1.2Co0.13Ni0.13Mn0.54O2 thin film cathode (or written as 0.55Li2MnO3–0.45LiCo1/3Ni1/3Mn1/3O2 based on mass ratio), is studied by using above-mentioned SPM techniques. This thin film cathode was deposited on commercial Si/Pt by Pulsed Laser Deposition (PLD) technique. The macroscopic electrochemical properties have been studied by Yan and colleagues.28
During the PLD deposition of the film, a KrF excimer laser beam (248 nm, 180 mJ) (Lambda Physik, USA) was used at a repetition frequency of 10 Hz. Thin films were deposited on Si/Pt substrates at 650 °C with an oxygen partial pressure of 350 mTorr. The target–substrate distance was kept at 20 mm during the film deposition. The as-deposited thin films were post-annealed at 800 °C with oxygen flow for 40 min.
Fig. 1(a) and (b) schematically shows the set-ups for the biased-AFM and c-AFM measurements used in this study. During the c-AFM measurement, the bias is applied through the bottom electrode (Pt layer on the substrate) whereas the SPM tip is grounded. On the other hand, for biased-AFM measurement, the bottom electrode is grounded and the bias is applied through the SPM tip. During biased-AFM, highly concentrated electric field corresponding to the applied bias can be induced at the tip–junction area. After bias application, volume changes of the cathode film are detected through the SPM tip. In the ambient air condition, a water meniscus may form in the tip/cathode junction, as shown in Fig. 1(c). This water meniscus may serve as a lithium reservoir which is composed of LiOH resulting from lithium and H2O reaction and renders the tip–electrode system reversibly.30 The bias applied through the effective contact region can promote lithium intercalation between the cathode thin film and the LiOH layer.30,31 In order to confirm the these effects of the ambient air, i.e., the possible water meniscus as lithium reservoir, the biased-AFM experiments were also conducted under synthetic air (H2O < 5 ppm (V)) and Ar gas (H2O < 0.01 ppm (V) and O2 < 0.02 ppm (V)), respectively. Overall, during the total process, charge conservation laws should be maintained both locally and globally.
In this work, the current maps have been obtained by c-AFM in the range of 0.5 V to 3 V with a step of 0.5 V over an area of 1.5 × 1.5 μm2. Further increasing the bias voltage, i.e., beyond 3.5 V, may cause deteriorated topography change in this material. This agrees well with the previous ESM (Electrochemical Strain Microscopy) study on the same cathode material, there was a significant change on topography between 3 and 4 V.29Fig. 2(a) shows the AFM topography image and Fig. 2(b) is the current image (under 3 V bias) overlaid on the topography image. There is a slight shift between the AFM and c-AFM images. Generally speaking, conductivity is defined as σ = (I/V)(l/A), where l is the distance between two points at which the voltage is measured and A is cross-sectional area perpendicular to the direction of the current. For a uniform material with a well-defined dimensions, the conductivity is proportional to the ratio of (I/V).10 In the c-AFM measurement of this study, the tip contact area and film thickness are generally fixed, also the external voltage is a constant through the measurement, hence the current variations can reflect the conductivity changes. The yellow-colored regions in current image correspond to locations with higher current (conductivity). Fig. 2(b) shows that the conductivity in most of the grain boundaries is much higher than that in the grain interiors. The current reduces further as the positions move away from grain boundaries. It is generally believed that the atoms located at grain boundaries are much more mobile than those inside the grains and hence the solute atoms can diffuse much faster along the grain boundary.35 The current differences between the grain interiors and boundaries are attributed to the differences in the Li-ions diffusion properties. As the grain boundaries can form a network within the material, therefore the electrochemical activity and diffusion of the Li-ions can be further enhanced at the grain boundaries. However, it is also noticed that not all the grain boundaries show higher current. For example, the current is lower at the boundaries of the five grains [marked by G1 to G5 in Fig. 2(a)]. This is probably because these grains are located at relatively higher position than other grains, as shown in the AFM topographic height images [Fig. 2(a)] where the bright regions correspond to the higher location. Therefore their grain boundaries may not be interconnected with other grains, hence, the conductivity of these grain boundaries are reduced.
Fig. 3 shows the changes of the current images with increasing bias from 0.5 V to 3 V. Again, the bright regions in current images correspond to the more conductive location. Within the range of 0.5 V to 1 V [Fig. 3(a) and (b)], the current can only be detected in few grains and most of grains show lower current. Further increasing the bias to 1.5 V [Fig. 3(c)], the conductive regions increase with certain grain boundaries showing high current value. When the bias reaches to 2 V [Fig. 3(d)], the current level of the grain boundaries increases abruptly. With further increasing the bias to 3 V [Fig. 3(d) to (f)], the current within the whole scanning area increases significantly. More importantly, high current position extends from grain boundaries to the grain interiors. Fig. 3(g) shows the comparison of current line section profiles along the lines in the images (d) to (f), respectively. Current between the grain interiors and boundaries changes from ∼1 nA to ∼5 nA when the bias changes from 2 V to 3 V. The width of the high current peaks also increases [Fig. 3(g)], this further confirms the extension of conductive regime from the grain boundaries to interiors.
This conductance characteristics in grains are further studied with the I–V measurement on the grains with different sizes. The tip is located at the center of the grains and the bias is applied with the same waveform which is shown in Fig. 4(b). Fig. 5(a) shows the topography of the investigated grains, where the four measured points are also shown. Fig. 5(b) shows the current as a function of time at the four locations during the voltage sweeps. As shown in Fig. 5(b), all grains show similar conductance change behaviors, except that the current level for the larger grains is smaller than that for the smaller grains at the same voltage. On the other hand, the current responses at different locations on one grain are further measured [Fig. 5(c) and (d)]. As shown in Fig. 5(d), when the tip is located near grain boundary (A to E), the conductance characteristics doesn't change much, almost independent of the tip locations. When the tip is located away from grain boundary (F to H), the current level becomes much lower, as shown in Fig. 5(d). Therefore, the results confirm that the conductance characteristics are closely dependent of the locations in a grain.
For comparison purposes, the positive biases are first applied at the grain interior for duration of 60 s. Fig. 7(a) to (c) are the deflection images scanned before applying the bias, immediately and 30 minutes after applying the +7 V bias at the same location. The biased position is located at the center of the red circle. Fig. 7(b) shows, after applying the +7 V bias at grain interior, several disperse protuberances emerge on the grain surface near the biased location. The features of other grains still remain the same as those before application of the bias. When positive bias is applied to the tip in biased-AFM, Li-ions are repelled away from the tip, resulting in a small region of depletion of Li-ions under the tip.36 The disperse protuberances indicate the evacuation of the Li-ions induced by positive bias since the c-axis of the lattice can expand as a result of increasing electrostatic repulsion between the TMO2 slabs due to the remove of the Li-ions.37 After 30 minutes, the dispersed protuberances disappear [Fig. 7(c)]. This indicates that Li-ions diffuse to the area with lower Li-ions concentration, resulting in the relaxation of Li-ions. In addition, although the chemical identification is not available due to the limitation of SPM, phase images still offer very useful information on the surface property and composition changes.16 Phase images show that the protuberances shift phase angle to a smaller value [Fig. S1, ESI†]. After 30 minutes, phase angle of the locations where protuberances once appear shift positively. This indicates the loss of Li-ions can shift the phase angle negatively. This result suggests a possible relationship between the Li-ions diffusion with the AFM phase images which need further investigations.
After this experiment, the bias was applied at selected grain boundaries region for duration of 60 s. Fig. 8(a) to (f) are the deflection images of the same area pre-biased, under 3 times of +7 V, followed by −7 V, and then +7 V bias again. The biased position is located at the gain boundary region (center of the red circle). Comparing with the pre-biased image, after applying +7 V bias for the first time at the grain boundary, those grains with flat facet (assuming with good crystallization), such as those marked by G1, G2 and G3, expand in a certain direction beyond the biased site, and it is found several steps emerge at certain grain boundaries [Fig. 8(b)]. On the other hand, the grains with round features such as that marked by G4, also beyond the biased site, but expand in all directions. The biased boundary only shows slightly morphological changes. When applying the +7 V bias for the second and third times, more and more steps begin to extrude along the same direction as those after the first bias [Fig. 8(c) and (d)]. Generally speaking, in the cathode with layered structure, the intercalation of Li-ions into cathode can cause an extension of the lattice in a-axis, while the de-intercalation of Li-ions out of the cathode can cause contraction of a-axis. In this work, since the cathode thin film is (003) orientated, the expanded grains and extruded steps (extension of a-axis) suggest that Li-ions move into these grains, causing the surface deformation, and this process should be similar to the intercalation process. Under the positive bias, lithium ions move into the cathode layer, hence causes the surface saturated with lithium ions and results in the volumetric expansion.31 With the application of the multiple times of biases, more Li-ions can be diffused into those grains. Compared with the observation where the bias is applied at grain interior, this result shows that Li-ions can diffuse into the surrounding grains through the interconnected grain boundaries. The boundary diffusion likely depends on the particular microstructure of the grain boundaries and the connectivity of the grains (or network). This result is consistent with the conductance map that shows certain grain boundaries can provide fast Li-ions diffusion pathways (Fig. 2). Additionally, Li-ions in the adjacent part can produce electrostatic repulsion to repulse those Li-ions driven by the tip bias, while the repulsive effect of electric field on Li-ions gradually decreases further away from tip–sample junction area. How far the Li-ions can be driven depends on the competitive electric field force and electrostatic repulsive force. Simulations and further experiment studies are still needed to provide more information on the interplay of electric field and electrostatic repulsion. In addition, such electrochemical behavior can be affected by the presence of a water meniscus that can both catalyze the electrochemical reaction and allow for increased ionic conduction. In order to confirm if above observation is caused by the water meniscus, the similar tests are performed in synthetic air and Ar gas (Fig. S2 and S3, ESI†). The results show that there is no deformation either in the biased position or surrounding area. These results confirm that the observed large deformation (Fig. 8) occur with the existing of water or moisture from the environment.
After applying the −7 V bias at the same location, the previously-formed extruded steps in unbiased area are vanished [Fig. 8(e)], and this is accompanied by the shrinkage of the grains (G1, G2 and G3). Note that the grain shape recovers to its original appearance but the grain size generally becomes smaller. Grain G4 also shrinks and becomes smaller. The shrinkage of the grains (contraction of a-axis) suggests Li-ions move out of these grains, and this process should be similar to the de-intercalation process. When the negative bias is applied, the “de-intercalation” occurs at the reservoir interface and volumetric shrinkage is induced.30,31 Therefore, the results in Fig. 8 suggest, when applying positive and followed by negative bias at the same location, Li-ions can reversibly “intercalated” into and “de-intercalated” out of the grains in the nearby locations. It is also noticed that the biased position [circled position in Fig. 8(e)] has been damaged after the application of the negative bias, whereas the nearby grains show reversible changes. There are some new and smaller nucleated particles at the biased location. This indicates the negative bias can induce extraction of Li-ions to the surface followed immediately by the precipitation of Li hydroxides and carbonates due to the direct reaction involved with Li and ambient moieties (H2O, O2, CO2, etc.).38,39 Furthermore, there are also some new grains formed beyond the biased location (marked by yellow arrows such as N1), and this indicates the possible exfoliation of the particles due to the “de-intercalation” of the Li-ions. Delamination is observed in grain G1 (marked by blue arrows), and it suggests non-uniform Li-ions “de-intercalation” in this grain. From a full-cell battery point of view, exfoliation and delamination due to non-uniform Li-ions diffusion may cause loss of contact between the cathode and electrolyte, and this can lead to the aging of the battery. Comparing with the results in synthetic air and Ar gas, again, there is no deformation observed at surface under negative bias (Fig. S2 and S3, ESI†). These results are also consistent with what reported on Li1.2Co0.13Ni0.13Mn0.54O2 particles.15
A bias of +7 V is then applied again at the same location, and the topography images are shown in Fig. 8(f). It is observed that most of the grains expand again and new steps are extruded from the grain structure, this indicates the “re-intercalation” of the Li-ions under the positive bias. However, the topography changes due to Li-ions “re-intercalation” are different from those caused by the first Li-ions “intercalation” process [Fig. 8(b)]. The extruded steps are not well arranged as those formed during the first Li-ions “intercalation” process. For example, the grain G2 expands again but becomes delaminated; grain G3 also expands but in a different direction; and the grain G4 expands but to a much smaller degree. The newly formed grain N1 (during the −7 V bias) does not disappear but expands obviously. The grain G1 still remains delaminated and does not recover to its initial state as an integrated grain. These unrecoverable topography changes imply the degradation and structural changes of the cathode material during the “intercalation” and “de-intercalation” processes. This experiment has provided a new perspective for understanding the underlying mechanisms of the poor cycling performance of the cathode in a full-cell battery system.
Furthermore, phase images and distribution histogram of phase angles (data fitted by using Gaussian function theory) show that the average phase angle shifts positively after Li-ions “intercalation” and negatively after Li-ions “de-intercalation” processes [Fig. S4, ESI†]. After 2nd Li-ion “intercalation” process, phase angle shift positively again. During the AFM measurement, phase angle shift was primarily originated from the difference in mechanical properties, such as elastic modulus, surface hardness and adhesion energy.16,40,41 Due to the complex interpretation of phase images, the precise explanation for the relationship between phase angle shift and Li-ion concentration is still a challenge issue and need to be studied further. However, this study indicate that the possible relationship between phase angle shift and Li-ion concentration. Since the characterization on Li-ion concentration is very significant for in situ SPM studies on the LIB materials, more efforts should be put to investigate the relationships between the phase values and Li-ion content in the future studies.
The biased-AFM results indicate that the Li-ion diffusion under the biases is different along the grain boundary and within the grain interior. More specifically, the bias applied at grain interior can only cause localized Li-ions diffusion inside the grain, whereas the bias applied at grain boundary can cause the Li-ions “intercalation” into and “de-intercalation” out of the surrounding grains reversibly. This is due to the fact that grain boundaries can form a continuous network in polycrystalline materials, and this network can provide a fast diffusion path for ion diffusion.
Moreover, this work has applied the positive and negative biases in sequence at a single-point to visualize the processes of the Li-ions move into and out, i.e., similar to the “intercalation” and “de-intercalation” processes, the layered structure in the cathode film without assembly of a full-cell battery system. In addition, exfoliation and delamination, degradation and structural changes of the cathode material are observed during the Li-ions movements. These observations at the nanoscopic scale offer valuable insights into the elucidation of aging mechanisms of cathode materials in a realistic battery device. This work also shows that the in situ characterization of the local electrochemical phenomena can afford clues to improve the performance of LIB systems.
Footnote |
† Electronic supplementary information (ESI) available: Biased-AFM phase images (bias is applied at grain interior) of the Li1.2Co0.13Ni0.13Mn0.54O2 cathode thin film; biased-AFM phase images (bias is applied at grain boundary) of the Li1.2Co0.13Ni0.13Mn0.54O2 cathode thin film. See DOI: 10.1039/c6ra17681j |
This journal is © The Royal Society of Chemistry 2016 |