Conversion Reaction of CoO Polycrystalline Thin Films Exposed to

We have studied the reaction of 5 nm thick polycrystalline CoO films with atomic lithium as a model for the discharge of lithium-ion conversion batter...
0 downloads 0 Views 3MB Size
Download PDF
Article pubs.acs.org/JPCC

Conversion Reaction of CoO Polycrystalline Thin Films Exposed to Atomic Lithium Ryan Thorpe,†,§ Sylvie Rangan,† Mahsa Sina,‡ Frederic Cosandey,‡ and Robert A. Bartynski*,† †

Department of Physics and Astronomy and Laboratory for Surface Modification, Rutgers University, 136 Frelinghuysen Road, Piscataway, New Jersey 08854, United States ‡ Department of Materials Science and Engineering, Rutgers University, 607 Taylor Road, Piscataway, New Jersey 08854, United States S Supporting Information *

ABSTRACT: We have studied the reaction of 5 nm thick polycrystalline CoO films with atomic lithium as a model for the discharge of lithium-ion conversion battery electrodes. The electronic structure has been investigated with X-ray photoemission, ultraviolet photoemission, and inverse photoemission, while the morphology, crystal structure, and unoccupied states of the films have been examined with transmission electron microscopy. It is found that exposure to atomic lithium leads, at room temperature, to partial conversion with formation of Co and Li2O, but also of a Li2O2 or LiOH overlayer at the surface of the sample. As full conversion was obtained at 150 °C, a comparison with roomtemperature measurements enables the understanding of the kinetic limitations during lithiation.

■

INTRODUCTION Lithium-ion conversion batteries represent a significant shift from the paradigm of conventional Li-ion battery operation. Rather than exploiting the intercalation of lithium ions in layered electrode materials such as LiCoO2 and LiC6, conversion electrode materials undergo a complete phase transformation upon exposure to Li.1−3 Transition metal compounds are particularly well suited as electrode materials in conversion cells as they can be fully reduced from Mx+ to M0 (in general x ≥ 2) upon exposure to lithium, thereby producing two or more electrons per transition metal ion. Metal fluorides such as FeF2 have exhibited sufficiently high voltages to warrant consideration as cathode materials, while metal oxides, sulfide, nitrides, and phosphides are candidate anode materials due to their low voltage.4−6 Additionally, polyanionic compounds such as FeOxFy have been studied as cathode materials due to their tunable electronic properties.7 The 3d transition metal oxides have attracted increasing attention from the battery community as potential anode materials due to their stability, natural abundance, and the many methods by which they can be synthesized in nanostructured morphologies.8−10 Their high gravimetric charge storage capacity and moderate voltage vs Li+/Li0 (0.2−1.4 V) make metal oxides attractive candidates for Liion anodes in applications where low voltage output and high charge density are required, e.g., mobile electronic devices.11,12 In a study by Badway and co-workers, a CoO anode cycled against a LiCoO2 cathode exhibited an output voltage of 2 V and a specific energy of 120 Wh/kg. That compared well to the specific energy of 180 Wh/kg obtained for similar cells with © 2013 American Chemical Society

LiC6 anodes but for which the output voltage is typically significantly larger.12 This is due to the fact that CoO can accommodate a larger lithium content than current LiC6 anodes. Indeed, CoO has exhibited capacities as high as 700 mAh/g at room temperature over 100 cycles, undergoing the following reversible reaction upon exposure to lithium13 2Li+ + 2e− + CoO → Li 2O + Co0

(1)

This represents a 2-fold increase in gravimetric charge storage capacity over current LiC6 electrodes.14−16 The increased charge density of conversion electrodes comes at the expense of structural stability. While the morphology of intercalation compounds is largely unchanged by the incorporation of lithium ions, conversion materials undergo a phase separation that results in a total structural reorganization of the electrode.5,6 This drastic change in morphology can lead to poor reaction kinetics and eventually capacity losses in an electrochemical cell.17 These issues motivate the need for fundamental research into the reaction pathways involving Li and CoO. To understand the phase progression during the conversion reaction of CoO, we have performed a study of high-purity CoO polycrystalline films grown in ultrahigh vacuum, sequentially exposed to atomic Li. The electronic structure of the pristine films and of the products of lithiation was studied using X-ray photoemission spectroscopy, UV photoemission Received: May 17, 2013 Revised: June 18, 2013 Published: June 18, 2013 14518

dx.doi.org/10.1021/jp404875g | J. Phys. Chem. C 2013, 117, 14518−14525

The Journal of Physical Chemistry C

Article

Figure 1. XPS survey scan of the initial CoO film after degassing. The corresponding Co 2p and O 1s core level spectra are added in the inset.

compensation using a combination of low-energy Ar+ ions and electrons was used during XPS measurements. The C 1s signal from adventitious contamination was chosen as a binding energy reference for all XPS spectra and was set at −284.8 eV. Shirley backgrounds were removed from each core level XPS spectrum presented in this work unless otherwise indicated. The UPS spectra and the IPS spectra were referenced to the Fermi level of a clean gold sample in contact with the sample. TEM. Bright-field TEM images, selected area electron diffraction (SAED) pattern, annual dark field STEM (ADFSTEM) images, and electron energy loss spectra (EELS) were obtained using a JEOL 2010F operated at 197 kV and equipped with a Gatan GIF 200 spectrometer. The EELS spectra were obtained with a collection half angle of 27 mrad and convergence angle of 10 mrad and with an energy resolution of 1.1 eV. EELS spectrum images (32 × 32 pixels) obtained with a 0.3 nm probe size and a 1.25 nm spatial resolution were collected to obtain information on the spatial distribution of Co species. Sample Preparation. For the spectroscopic measurements, 5 nm thick CoO films were prepared by e-beam evaporation of Co metal in a 1 × 10−7 Torr background O2 pressure, onto clean copper foil at room temperature. Copper was chosen as an inert, nonreducing substrate, compatible with the annealing temperatures used in this study. For TEM measurements, films were grown on thoroughly degassed SiOxNy TEM membranes from SiMPore. The films were grown with a typical deposition rate of ∼1 Å/min as measured by a quartz crystal microbalance, and the final thickness was confirmed ex situ by Rutherford backscattering spectrometry. As-grown samples were briefly exposed to air during transfer to the ESCALAB instrument or to the TEM. To remove surface contaminants after transfer in the atmosphere, the films were degassed at 300 °C for 5 min in the ESCALAB. XPS was used to confirm the purity and stoichiometry of the resulting CoO films. Lithiation was performed in situ using a Li getter source (SAES Getters) while maintaining the CoO films either at 25 or 150 °C. TEM samples were not degassed after the short exposure to air, but the effects of surface contamination using this bulk sensitive technique are expected to be negligible.

spectroscopy, and inverse photoemission spectroscopy. The crystal structure and film reorganization were examined in parallel with transmission electron microscopy. By avoiding the use of electrolytes, separators, and packaging materials, we aim to observe the intrinsic properties of reaction between the active conversion materials. It is also important to note that our solid state chemical conversion reaction differs from typical electrochemical reactions since electrons do not flow through an external circuit but instead arrive with the atomic Li at the surface of the active material. Although the chemical conversion of CoO upon lithium exposure was readily observed at room temperature for low Li dose, it is found that kinetic effects related to lithium diffusion in lithium oxide phases, that develop upon continued lithiation, result in significant differences in the rate of the chemical conversion process at 25 °C as compared to 150 °C. These results contrast with the facile conversion of FeF2 with lithium exposure and subsequent formation of LiF and Fe, obtained previously using the same approach.4 The results presented here are relevant for conversion batteries based on various binary transition metal oxides, as the reaction products are expected to be similar for these materials. Moreover, these results might have significant implications for Li-air batteries, where the formation of lithium oxide phases and subsequent kinetic limitations are key issues.18−23

■

EXPERIMENTAL TECHNIQUES XPS−UPS−IPS. Most X-ray photoemission spectroscopy (XPS) and most ultraviolet photoemission spectroscopy (UPS) measurements presented here were performed using a Thermo Scientific ESCALAB 250Xi with a base pressure less than 1 × 10−9 Torr. Core level photoelectrons were excited with X-rays from a monochromated Al Kα source with an energy of 1486.6 eV (∼0.5 eV instrumental resolution) and valence band photoelectrons by a He II line with an energy of 40.8 eV (∼0.1 eV instrumental resolution). The photoelectron spectra were measured in angle-integrated modes. Additional electron affinity and band edge measurements were performed in a separate chamber housing UPS and inverse photoemission spectroscopy (IPS) with respective resolutions of 0.3 and 0.6 eV.24 The thin CoO films were conductive enough for spectroscopic measurements. However, to prevent possible charging effects of the products of the lithiation, charge

■

RESULTS AND DISCUSSION Characterization of the CoO Polycrystalline Films. Before discussing the conversion reaction itself, it is useful to 14519

dx.doi.org/10.1021/jp404875g | J. Phys. Chem. C 2013, 117, 14518−14525

The Journal of Physical Chemistry C

Article

consider the electronic structure of the initial CoO films. In this work, both occupied and unoccupied states have been measured on CoO films for which both the crystal structure and morphology have been characterized. Figure 1 shows an XPS survey spectrum of the CoO film prior to lithiation, with a carbon contribution of less than 10% of the total atomic species confirming the cleanliness of the film. The Co 2p core level spectrum, presented in the inset of Figure 1, consists of several structures corresponding to crystal field multiplets (the two main peaks are found at −780.6 and −796.5 eV) and shakeup/shake-off satellites components, indicative of stoichiometric CoO.16,25−28 It should be noted that no metallic Co component is visible in this spectrum, indicating the presence of only one oxidation state, Co2+. The corresponding initial O 1s spectrum (inset in Figure 1) consists of a main peak at −530.0 eV attributed to O2− anions in the lattice and of a smaller peak at −531.9 eV attributed to adsorbed hydroxyl species or defect structures at grain boundaries.28 The relative intensities of Co and O in this sample indicate a Co:O ratio of 1:1. The valence and conduction bands measured, respectively, using UPS and IPS are presented in Figure 2. The zero of

which the transition originates from the Co 2p level reflect a density of state weighted by the extent of hybridization between Co 3d or Co 4s states and O 2p or O 3s states. Similarly, spectra measured at the O 1s-edge selectively probe states of the conduction band hybridized with O 2p states. In IPS, the unoccupied states of all orbital character are probed only weighted by the cross section of the electronic states involved in the inverse photoemission process. In the case of CoO, the unoccupied electronic structure has been described to first order in terms of Co−O hybridized states arranged as follows with increasing energy: Co 3d(t2g)−O 2p < Co 3d(eg)−O 2p < Co 4s−O 2p < Co 4p−O 2p.35 Thus, using the incomplete but complementary information provided by EELS or XAS at the Co 2p and O 1s edges, the conduction band spectrum measured in IPS can be understood as follows: unoccupied states from 0 to 10 eV are mainly due to hybridized Co 3d states, while from 10 to 15 eV are found Co 4s−O 2p states, followed by Co 4p−O 2p states above 15 eV. As the valence band and the conduction band spectra are both referenced to the Fermi level of the CoO sample, the separation between the band edges is related to the electronic transport gap. By linearly extrapolating the band edges to the background of the spectra, a valence band maximum of −1.0 ± 0.1 eV and a conduction band minimum of 1.1 ± 0.1 eV can be defined. Using this method, a transport gap of 2.1 ± 0.2 eV is measured. Similar results have been obtained earlier from a combination of XPS and BIS measurements, although these measurements suffered from a larger experimental broadening.26 Finally, we report the electron affinity (EA) of the CoO film. This quantity is obtained as the energy difference between the vacuum level and the conduction band minimum, measured from a combination of UPS and IPS spectroscopies. A wide valence band spectrum generated from a He I radiation (hν = 21.2 eV) is measured with a −5 V applied bias on the sample. The total width of the spectrum, W = 16.1 eV, is extracted using a linear extrapolation of the data to the background intensity level at both the high- and low-kinetic-energy ends of the spectrum. The electron affinity is obtained as EA = hν − W − Egap, where hν is the photon source energy and Egap = 2.1 eV is the experimentally measured transport gap from UPS and IPS spectroscopies. Using this procedure, we arrive at an electron affinity of EA = 3.0 eV. TEM measurements summarized in Figure 3 have been performed to characterize both the morphology and the crystalline structure of the CoO films. The pristine films have been measured as-grown and after the degassing procedure to ensure that no changes occurred upon annealing in ultrahigh vacuum (UHV). Small-area XPS was also performed on the TEM membrane-supported films for consistency. Figures 3(a) and 3(b) show the ADF-STEM image and the corresponding SAED pattern, respectively, for a CoO film grown on a SiOxNy membrane. The SAED pattern confirms the presence of the cubic Fm3̅m phase of CoO. In ADF-STEM mode, high-Z elements such as Co will appear as bright features. This interpretation is supported by smaller-scale STEM images (Figure 3(c)) and the corresponding elemental maps for Co and Si (shown, respectively, in Figure 3(d) and (e)). These maps were obtained from the Co-M edge and Si-L edge EELS spectral intensity, acquired at each point of the image. In these images, Co-containing particles appear on a Sirich background, attributed to the TEM thin membrane material. The CoO film thus appears as interconnected particles

Figure 2. Valence band spectrum measured in UPS (energy < 0) and conduction band spectrum measured in IPS (energy > 0) of a 5 nm CoO film on Cu foil. A linear extrapolation of the band edges leads to a measure of the transport gap of CoO of 2.1 eV. The energy scale is referenced to the Fermi level of the system.

energy for each spectrum is chosen as the Fermi level of the system, so that occupied states have a negative energy and the unoccupied states a positive energy. In this figure, the conduction band is a composite spectrum obtained for different primary energy of electrons (E = 20.3 eV to image the region from 0 to 6 eV and up to E = 34.3 eV for higher lying unoccupied states).29 The valence band was measured using a photon energy of 40.8 eV. The valence band spectrum of Figure 2 is consistent with those of cleaved CoO(100) crystals.30−32 The electronic states with energies between the valence band maximum and −3 eV are attributed to Co 3d-derived states, while the intensity from −3 to −8 eV is primarily due to O 2p-derived electronic states.33,34 The last visible structure found at −10 eV is attributed to an unscreened Co 3d-like satellite.32 The IPS spectrum of Figure 2 can be compared to the unoccupied states measured previously using electron energy loss spectroscopy (EELS),35 X-ray absorption spectroscopy (XAS),36 and bremsstrahlung isochromat spectroscopy (BIS).26 The electronic transitions generated in EELS and XAS are both governed by dipolar selection rules, resulting in a partial description of the unoccupied density of states. Spectra for 14520

dx.doi.org/10.1021/jp404875g | J. Phys. Chem. C 2013, 117, 14518−14525

The Journal of Physical Chemistry C

Article

and −793.5 eV, corresponding to to the 2p3/2 and 2p1/2 states in metallic Co.25 A reference spectrum from a clean Co foil, shown at the top of Figure 4(b), confirms these peak assignments. The metallic contribution to the Co 2p core level spectra increases with Li exposure. Concurrently, the features associated with Co2+ decrease, indicating an almost total conversion from Co2+ to Co0. In this figure, the Co 2p peaks have been normalized by height to enhance the visual differences between spectra. Without this normalization, the Co 2p signal is strongly attenuated by more than two-thirds from its initial intensity at the end of lithiation, indicating the presence of a thick overlayer. In parallel, the O 1s core level spectrum shown in Figure 4(c) exhibits upon initial lithiation a slight broadening, followed by a shift to a lower binding energy as a function of Li exposure, indicating the formation of Li2O at the expense of CoO.16 After the final lithiation, the main O 1s peak energy is −529.6 eV. A smaller feature, that could be associated with either Li2O2 or LiOH, also appears at −532.0 eV immediately after the first Li exposure. This peak grows more intense and shifts toward higher binding energy after subsequent Li exposures. The final peak position of this component is −532.4 eV, consistent with previous XPS studies of lithium peroxide formation.37−39 Figure 4(d) shows the corresponding UPS valence band spectra before and after sequential Li exposures. The line shape of the clean CoO spectrum was described in the previous section. After the first lithiation, two new features appear at binding energies of −6.4 and −10.4 eV, generally attributed to O 2p states in Li2O2. However, in the absence of clear experimental reference spectra, it is difficult to categorically reject the presence of LiOH.38,39 These peaks then develop and shift toward more negative binding energy with increasing Li exposure to reach −6.6 and −10.9 eV, respectively. A smaller shoulder at −5.0 eV, corresponding to O 2p states in Li2O, is also observed and increases with Li exposure.38,39 Lastly, starting from the well-defined band edge of CoO, a Fermi edge develops with increasing Li exposure, consistent with the formation of metallic Co, as shown in Figure 4(e). It should be noted that due to the particular photon energy used in this measurement (40.8 eV) the valence band spectra are much more surface sensitive than the XPS core level spectra. The strong contribution to the electronic structure of the Li2O2 bands in the valence band indicates that Li2O2 is most likely forming an overlayer on top of Li2O or Co. Assuming that this is the case, a simple model using the XPS intensities of the last lithiation step of Figure 4, and assuming a 25 Å attenuation length in the lithium oxides compounds, one can estimate the thickness of Li2O2 to be 5 Å (see Section S3, Supporting Information). Figure 5 shows the results of a TEM study of the morphology and structure of the final product of the lithiation reaction of a CoO film grown on a SiOxNy membrane and fully reduced by exposure to Li at 150 °C in UHV (as confirmed by small area XPS). A large area ADF-STEM image and a SAED pattern from the resulting film are shown, respectively, in Figures 5(a) and 5(b). A smaller area image, with its corresponding elemental map at the Co-M and Si-L edges, is also shown in Figures 5(c), 5(d), and 5(e), respectively. The STEM images indicate a global reduction of the Co-containing particle size, with a distribution centered around 2−3 nm (as shown in the diagram of Figure 5(f)), consistent with electrochemical conversion studies of CoO.13 The final particles appear interconnected, as expected from a conversion

Figure 3. (a) ADF-STEM image of a CoO thin film on a SiOxNy membrane. (b) SAED pattern confirming the cubic structure of CoO (Fm3m ̅ ). (c) STEM image and corresponding elemental maps at (d) the Co-M and (e) the Si-L edges. (f) Size distribution of the CoO nanoparticles. (g) Small area Co 2p spectrum measured on the same film.

with an average size distribution centered around 5 nm (as shown in Figure 3(f)). Finally, the small area XPS Co 2p spectrum measured on the same film confirms the presence of CoO. Lithiation of the CoO Films. Having established the quality of the CoO polycrystalline thin films, both on the Cu substrate and on the SiOxNy membrane, lithiations of the films were performed in UHV at a chosen temperature of 150 °C. This temperature was low enough to prevent CoO reduction but high enough to enhance the diffusion rate of Li in the sample.18 This aspect will be discussed in more detail in the next section. Figure 4 shows the evolution of the core level and valence band spectra, respectively, measured by XPS and UPS, of a CoO film grown on a Cu foil, before and after sequential exposure to Li. Although suffering from a low photoemission cross section at this photon energy, a spectral feature from the Li 1s core level (shown in Figure 4(a)) can be observed above the background even at the lowest Li exposures. The peak intensity increases after each sequential Li exposure. The broad shape of the Li 1s spectra does not allow an absolute decomposition of the chemical species present but is consistent with a contribution from a main peak at −54.7 eV corresponding to Li2O and a smaller peak at −55.5 eV attributed to Li2O2 or LiOH.16,21 This deconvolution scheme will be justified by the analysis of the remaining core levels and by XPS spectra of lithiated Cu, as shown in Figure S5 (Supporting Information). Upon exposure to Li, new features appear in the Co 2p spectra (shown in Figure 4(b)) at binding energies of −778.4 14521

dx.doi.org/10.1021/jp404875g | J. Phys. Chem. C 2013, 117, 14518−14525

The Journal of Physical Chemistry C

Article

Figure 4. Evolution of the (a) Li 1s, (b) Co 2p, and (c) O 1s core level spectra and of the (d) valence band spectra before and after sequential Li depositions. (e) A zoomed-in version of the Fermi edge region is added to emphasize the metallic edge development. Green arrows indicate increasing Li exposure. The topmost Co 2p spectrum was obtained from a clean polycrystalline Co sample and is used as a reference for metallic Co. The Li 1s spectra have been normalized to the Co amount (using the Co 3p intensity). The Co 2p and O 1s spectra have been normalized by height to accentuate visual differences between lineshapes. Valence band spectra have been normalized to the highest binding energy intensity.

material able to sustain cyclability.40 The small area XPS Co 2p spectrum shown in Figure 5(g) is similar to the one observed at the final lithiation step of Figure 4 and indicates that a large amount of the Co is now in the metallic state. An analysis of the SAED pattern confirms the presence of a metallic Co phase (characterized by a P63/mmc symmetry) and of Li2O (of cubic Fm3̅m structure) coexisting with a minor component of CoO (of cubic Fm3̅m structure). This latter phase could be unreacted CoO but is most likely due to oxidation occurring during the brief exposure of metallic Co to the atmosphere upon transport to the TEM. From these measurements, it is clear that upon lithiation the CoO film is reduced to a metallic form, with concomitant formation of at least two lithium oxides phase, Li2O and Li2O2 (although LiOH cannot be excluded). Interestingly, these findings indicate that lithiation of a pristine CoO film under UHV conditions drives the conversion reaction to the same final products and morphology observed in electrochemical cells upon lithium ion insertion in a CoO matrix. In this experiment, however, the sample temperature was maintained at 150 °C during exposure to lithium. In the following section, a comparison of the lithiation products of CoO at 150 and 25 °C will help in understanding some mechanistic aspects of the lithiation process. Quantitative Analysis: Conversion Reaction and Mechanistics. A semiquantitative analysis of the XPS data is necessary to further explore the conversion reaction mechanism. The relative concentrations of the three species Li, Co metal (Co0 oxidation state), and CoO (Co2+ oxidation state) can be obtained by analyzing the Li 1s and Co 2p core level

spectra shown in Figures 6(a) and 6(b). With regard to lithium, the Li 1s core level peak was sufficiently well separated from the Co 3p peak to allow the subtraction of a Shirley background and integration of the intensity over the relevant energy range (Figure 6(a)). Due to the absence of well-defined structure in the Li 1s photoemission peak, this intensity contains the contribution from both the Li2O and Li2O2 (or LiOH) phases. In the case of Co 2p, each Co 2p spectrum could be described by a linear combination of Co metal and CoO reference spectra (as shown in Figure 6(b)). This is indicative that no other Co oxidation state was present during lithiation. It is now possible to evaluate the extent of CoO reduction as a function of lithium exposure at 150 and 25 °C, by comparing a set of XPS data similar to that presented in Figure 4 but lithiated at room temperature. In Figure 7, the relative fractions of cobalt in the Co0 and Co2+ oxidation states have been plotted as a function of a normalized detected Li coverage, that is, the total Li 1s XPS intensity divided by the total Co 3p XPS intensity.41 A first look at Figure 7 indicates drastic differences due to the sample temperature during Li deposition. At 150 °C, the reduction of CoO into Co follows a linear trend and reaches about 80%. Near full conversion of similar samples was obtained with higher lithium exposures (shown in Figures S1 and S2, Supporting Information). At 25 °C, the sample exhibits a rapid initial transformation, but after a Li:Co ratio of 1, CoO reduction is significantly impeded, slowly approaching saturation to a value less than 50% with additional Li exposures. Another noticeable point is that the Li:Co ratio on the absissa axis of Figure 7 is greater than the one expected for a simple conversion reaction where two Li atoms are necessary to fully 14522

dx.doi.org/10.1021/jp404875g | J. Phys. Chem. C 2013, 117, 14518−14525

The Journal of Physical Chemistry C

Article

on clean metallic substrates such as copper exposed to Li. The formation of these compounds is the result of the high reactivity of metallic lithium with the residual O2 and H2O partial pressures of the UHV environment (with a base pressure of less than 1 × 10−9 Torr). Consequently, not all of the deposited Li is available to react with CoO. The formation of a Li2O2 layer (estimated to be 5 Å thick) is key to understanding the smaller extent of the CoO reduction at 25 °C. Raising the temperature to 150 °C did not significantly modify the amount of Li2O2 observed but rather could have improved Li diffusion through the Li2O2 barrier, thereby providing a source of Li that can react with the underlying CoO layer. From a thermodynamic perspective, the reduction of CoO by Li is highly favorable (ΔGf = −1.8 eV per Li), and the existence of an energy barrier to this reaction is unlikely at room temperature as electrochemical conversion readily occurs. Indeed it is observed here that lithiations made at 25 °C and at 150 °C induce the same amount of conversion, only for the first two or three sequential exposures. Only afterward, when a Li2O2 (or LiOH) layer develops, the extent of CoO reduction is greatly affected as a function of temperature. A similar temperature dependence has been observed experimentally in Li-air batteries, where temperature increase considerably improves rate and capacity. This was attributed to an increased permeation of Li ions in Li2O2.18 Consistent with this idea, lithium diffusion in Li2O2 is believed to be mediated by Li vacancies and is now thought to be the limiting factor in Li−O2 batteries.22,23 The presence of a LiOH coating at the surface of porous cathodes in Li−O2 cells was also reported from X-ray diffraction and was correlated with capacity fading.42 Once again, signatures of LiOH and Li2O2 are not sufficiently characterized to clearly identify these compounds with our present experimental techniques. Thermodynamic considerations, however, could favor LiOH: in the absence of a substantial energy barrier, the reaction of Li2O2 with water should be energetically favorable (ΔGf = −0.45 eV per Li) and could lead to LiOH formation. Surprisingly, Li2O2 formation has not been reported in conversion batteries using CoO as an electrode.14−16 More precisely, the lack of structural information in the electrochemical studies involving CoO does not rule out the presence of Li2O2 or LiOH in these systems. It could be argued that the stability of these lithium compounds in an electrochemical cell

Figure 5. (a) ADF-STEM image of a CoO thin film on a SiOxNy membrane lithiated at 150 °C until full CoO reduction. (b) SAED pattern revealing the presence of a mixed phase composed of Co (P63/mmc symmetry), CoO (Fm3̅m symmetry), and Li2O (Fm3̅m symmetry). (c) STEM image and corresponding elemental maps at (d) the Co-M and (e) the Si-L edges. (f) Size distribution of nanoparticles composing the final product. (g) Small area Co 2p spectrum measured on the same film.

reduce one CoO unit. This can be explained in part by the exponentially decreasing contribution of emitted photoelectrons as a function of the depth from which they originate below the surface. If Li2O2 or LiOH forms a film at the surface of the sample, the measured Li:Co ratio will be greater than the actual ratio of atoms in the film. The formation of these species in the UHV chamber was also observed in the absence of a CoO film (shown in Figure S4, Supporting Information), e.g.,

Figure 6. (a) XPS core level spectra of the Li 1s and Co 3p regions of the final product of the lithiation of a CoO film, showing the Shirley background used to extract the Li content of the samples. (b) Co 2p spectrum of a CoO film after several Li depositions and peak fit using a linear combination of CoO and Co reference spectra. 14523

dx.doi.org/10.1021/jp404875g | J. Phys. Chem. C 2013, 117, 14518−14525

The Journal of Physical Chemistry C

Article

Figure 7. Relative amounts of Co0 and Co2+ vs Li−Co ratio as measured by XPS for a CoO film lithiated at (a) 150 °C and (b) 25 °C.

by the low lithium mobility in these oxides.23 Full conversion was obtained at 150 °C, when presumably Li diffusion is activated in the surface oxides. The results presented here raise important considerations for conversion batteries that utilize transition metal oxide electrodes such as CoO, Cr2O3, Fe2O3, or FeOF. The formation of Li2O2 or LiOH compounds has not yet been reported in electrochemical studies of metal oxide batteries, but their presence cannot be ruled out. Hence, the poor cycling stability and voltage hysteresis in these materials could be due in part to the irreversible formation of Li2O2. Additional lithiation studies on thin epitaxial CoO films using angle-resolved XPS are currently underway to effectively probe the layered structure of the lithiated films. Moreover, a definitive characterization of lithium oxide compounds in XPS and UPS would be of great help to unravel the complexity of these systems.

environment could be modified in the combined presence of an electrolyte and of Co or CoO.14 It is important to note that exposing another conversion material, FeF2, to atomic lithium in conditions essentially identical to those used in the study reported here led to the formation of metallic Fe and LiF as final products.4 Li oxides appeared only after the full reduction of FeF2, when presumably excess metallic lithium was present at the surface of the film (shown in Figure S3, Supporting Information). There is thus a fundamental difference in lithium reactivity that could be explained kinetically by a slower reaction rate for CoO reduction competing with direct lithium oxidation by the residual atmosphere, accentuated by the reduced penetration depth of Li after formation of Li2O2 or LiOH at the surface of the sample, further increasing Li oxidation. Isotope exchange experiments might be very useful in sorting out these different possibilities. From these measurements, some qualitative kinetic aspects of the phase transformation during the conversion reaction of CoO are retrieved. As no intermediate Co+ oxidation state is observed during the conversion reaction, Li mobility in CoO is expected to be sufficiently large to allow access to at least two Li per Co2+ site. Another consequence of the high mobility of lithium is the morphology disruption: smaller interconnected Co particles could be the result of Co precipitation during the conversion process. Similar nanoparticle size reduction has been observed in CoO−Li electrochemical cells.13 In this respect, these results are again analogous to those observed during the conversion reaction of FeF2.4

■

ASSOCIATED CONTENT

S Supporting Information *

The calculation of the atomic percentages of the initial CoO film and the thickness of the Li2O2/LiOH overlayer are reported in the Supporting Information. XPS spectra from the full conversion of a CoO film and the formation of Li2O2/ LiOH on other substrates are also presented. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

■

*E-mail: [email protected].

CONCLUSION Thin films of polycrystalline CoO were grown by reactive deposition of Co metal onto Cu foil or SiOxNy membranes in 1 × 10−7 Torr O2 at room temperature. The overall quality of the films was probed using XPS and TEM. Additionally, using a combination of UPS and IPS measurements, the valence and conduction band spectra, as well as a transport gap of 2.1 eV and an electronic affinity of 3.0 eV were reported. Exposure to atomic lithium can lead to full conversion with formation of Co and Li2O, but the presence of a parasitic growth of Li2O2 or LiOH at the surface of the sample can suppress CoO reduction by impeding Li diffusion. The lithium oxide compounds forming an overlayer could be the result of the slower kinetics of CoO conversion (as opposed to what is observed for FeF24), leaving metallic lithium free to react with the residual gases in UHV. This process could be exacerbated

Notes

The authors declare no competing financial interest. § Nanotechnology for Clean Energy IGERT.

■

ACKNOWLEDGMENTS This work was supported by the Northeastern Center for Chemical Energy Storage, an Energy Frontier Research Center funded by the U.S. DOE, BES under award No. DESC0001294. R.T. was supported by National Science Foundation Grant No. 0903661: Nanotechnology for Clean Energy IGERT.

■

REFERENCES

(1) Whittingham, M. S. Chem. Rev. 2004, 104, 4271−4301. (2) Goodenough, J. B.; Kim, Y. Chem. Mater. 2010, 22, 587−603. (3) Tarascon, J.-M. Philos. Trans. R. Soc., A 2010, 368, 3227−3241.

14524

dx.doi.org/10.1021/jp404875g | J. Phys. Chem. C 2013, 117, 14518−14525

The Journal of Physical Chemistry C

Article

(40) Wang, F.; et al. J. Am. Chem. Soc. 2011, 133, 18828−18836. (41) Several assumptions have been made to draw quantitative conclusions from these data. The first assumption is that the partially reacted film was a homogeneous mixture of Li2O, CoO, and Co0. Since XPS is exponentially more sensitive to surface layers than subsurface layers (due to photoelectron attenuation in the material), species that are preferentially found at the surface will appear much more intense than those in the bulk. This implies that the relative amount of Li deposited was likely overestimated by XPS intensities. The second assumption is that the Li 1s and Co 2p XPS spectra sampled the same depth in the sample, whereas the mean free path of electrons in solids depends heavily on their kinetic energy, increasing with electron energy in the energy range of XPS.43 Since the kinetic energy of Li 1s photoelectrons is approximately twice that of Co 2p electrons, XPS will again tend to overestimate the amount of Li in a sample. Despite these limitations, XPS analysis allowed for direct and reproducible comparison of CoO films lithiated at different temperatures. (42) Black, R.; Oh, S. H.; Lee, J.-H.; Yim, T.; Adams, B.; Nazar, L. F. J. Am. Chem. Soc. 2012, 134, 2902−2905. (43) Zangwill, A. Physics at Surfaces; Cambridge University Press: New York, 1988.

(4) Rangan, S.; Thorpe, R.; Bartynski, R. A.; Sina, M.; Cosandey, F.; Celik, O.; Mastrigiovanni, D. D. T. J. Phys. Chem. C 2012, 116, 10498−10503. (5) Amatucci, G.; Pereira, N. J. Fluorine Chem. 2007, 128, 243−262. (6) Cabana, J.; Monconduit, L.; Larcher, D.; Palacn, M. R. Adv. Energy Mater. 2010, 22, E170−E192. (7) Pereira, N.; Badway, F.; Wartelsky, M.; Gunn, S.; Amatucci, G. J. Electrochem. Soc. 2009, 156, A407−A416. (8) Do, J.-S.; Weng, C.-H. J. Power Sources 2005, 146, 482−486. (9) Yu, Y.; Chen, C.-H.; Shui, J.-L.; Xie, S. Angew. Chem., Int. Ed. 2005, 44, 7085−7089. (10) Yao, W.; Yang, J.; Nuli, Y. J. Electrochem. Soc. 2008, 155, A903− A908. (11) Li, H.; Balaya, P.; Maier, J. J. Electrochem. Soc. 2004, 151, A1878−A1885. (12) Badway, F.; Plitz, I.; Grugeon, S.; Laruelle, S.; Gozdz, A.; Tarascon, J.-M. Electrochem. Solid-State Lett. 2002, 5, A115−A118. (13) Poizot, P.; Laruelle, S.; Dupont, L.; Tarascon, J.-M. Nature 2000, 407, 496−499. (14) Dolle, M.; Poizot, P.; Dupont, L.; Tarascon, J.-M. Electrochem. Solid-State Lett. 2002, 5, A18−A21. (15) Grugeon, S.; Laruelle, S.; Dupont, L.; Tarascon, J.-M. Solid State Sci. 2003, 5, 895−904. (16) Dedryvere, R.; Laruelle, S.; Grugeon, S.; Poizot, P.; Gonbeau, D.; Tarascon, J.-M. Chem. Mater. 2004, 16, 1056−1061. (17) Malini, R.; Uma, U.; Sheela, T.; Ganesan, M.; Renganathan, N. G. Ionics 2009, 15, 301−307. (18) Das, S. K.; Xu, S.; Emwas, A.-D.; Lu, Y. Y.; Srivastava, S.; Archer, L. Energy Environ. Sci. 2012, 5, 8927−8931. (19) Ogasawara, T.; Debart, A.; Holzapfel, M.; Novak, P.; Bruce, P. J. Am. Chem. Soc. 2006, 128, 1390−1393. (20) Christenson, J.; Albertus, P.; Sanchez-Carrera, R.; Lohmann, T.; Kozinsky, B.; Liedtke, R.; Ahmed, J.; Kojic, A. J. Electrochem. Soc. 2012, 159, R1−R30. (21) Lu, Y.-C.; Crumlin, E.; Veith, G.; Harding, J.; Mutoro, E.; Baggetto, L.; Dudney, N.; Liu, Z.; Shao-Horn, Y. Sci. Rep. 2012, 2, 1− 6. (22) Ong, S. P.; Mo, Y.; Ceder, G. Phys. Rev. B 2012, 85, 1−5. (23) Radin, M.; Rodriguez, J.; Siegel, D. Proc. Battery Congress 2011, 60, 1−6. (24) Rangan, S.; Coh, S.; Thorpe, R.; Bartynski, R. A.; Rochford, J.; Galoppini, E. J. Phys. Chem. C 2010, 114, 1139−1147. (25) Grosvenor, A.; Kobe, B.; Biesinger, M.; McIntyre, N. Surf. Interface Anal. 2004, 36, 1564−1574. (26) van Elp, J.; Eskes, W.; Kuiper, P.; Sawatzky, G.; de Groot, F.; Turner, T. Phys. Rev. B 1991, 44, 6090−6103. (27) Okada, K.; Kotani, A. J. Phys. Soc. Jpn. 1992, 61, 449−453. (28) Kim, K. S. Phys. Rev. B 1975, 11, 2177−2185. (29) For such a small energy span (Ep = 20.3−34.3 eV), variations of the cross section of the IPS process are neglected. (30) Thornton, G.; Brookes, N.; L.-Law, D.; Warburton, D.; Wincott, P. Phys. Scr. 1990, 41, 625−628. (31) Mackay, J.; Henrich, V. Phys. Rev. B 1989, 39, 6156−6168. (32) Shen, Z.-X.; Allen, J.; Lindberg, P. Phys. Rev. B 1990, 42, 1817− 1828. (33) Rödl, C.; Fuchs, F.; Furthmüller, J.; Bechstedt, F. Phys. Rev. B 2009, 79, 1−8. (34) Eder, R. Phys. Rev. B 2008, 78, 1−15. (35) Mitterbauer, C.; Kothleitner, G.; Grogger, W.; Zandbergen, H.; Freitag, B.; Tiemeijer, P.; Hofer, F. Ultramicroscopy 2003, 96, 469− 480. (36) Kurmaev, E.; Wilks, R.; Moewes, A.; Finkelstein, L.; Shamin, S.; Kunes, J. Phys. Rev. B 2008, 77, 165127. (37) Wu, Q.-H.; Thissen, A.; Jaegermann, W. Appl. Surf. Sci. 2005, 250, 57−62. (38) Wu, Q.-H.; Thissen, A.; Jaegermann, W. Surf. Sci. 2005, 578, 203−212. (39) Ensling, D.; Thissen, A.; Jaegermann, W. Appl. Surf. Sci. 2008, 255, 2517−2523. 14525

dx.doi.org/10.1021/jp404875g | J. Phys. Chem. C 2013, 117, 14518−14525