Correlation of Electrochemical Performance with Lithium

Jun 26, 2015 - Solid-state NMR studies of chemical exchange in ion conductors for alternative energy applications. Danielle L. Smiley , Gillian R. Gow...
1 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/JPCC

Correlation of Electrochemical Performance with Lithium Environments and Cation Dynamics in Li2(Mn1−yFey)P2O7 using 6Li Solid-State NMR Danielle L. Smiley, Matteo Z. Tessaro, X. He, and Gillian R. Goward* Department of Chemistry & Chemical Biology, and Brockhouse Institute for Materials Research, McMaster University, 1280 Main St. West, Hamilton, ON L8S 4M1, Canada S Supporting Information *

ABSTRACT: 6Li solid-state nuclear magnetic resonance (ssNMR) is used here to evaluate a series of Li2Mn1−yFeyP2O7 cathode materials in an effort to quantify ion exchange rates and diffusion pathways. Magic angle spinning (MAS) NMR of the series of mixed metal pyrophosphates reveals a trade-off between electrochemical performance and well-resolved NMR spectra resulting from the change in electronic structure of the transition metal redox center. In addition, 1D 6Li selective inversion NMR is employed to characterize Li ion dynamics in the fully Mn substituted member of the pyrophosphate series, where three of the four unique Li resonances are well resolved and labeled AB, C, and D, with AB corresponding to Li ions within one tunnel, and C and D Li ions residing in another. Despite limited inversion efficiency it is found that the utility of this experiment is not compromised so long as the initial magnetization conditions are well-defined. Initial fitting procedures involved the inclusion of all possible exchange pairs, a process which gave rise to consistently negative rate constants for C−AB or D−AB exchange, suggesting negligible exchange between these Li ions. Upon limiting the exchange model to ion exchange processes between the pairs of high and low frequency sites, rate constants of 45 ± 25 and 100 ± 30 Hz were obtained for C−D exchange at room temperature and 350 K respectively. Ion exchange pathways that are revealed by the exchange experiments imply limited mobility across distinct two-dimensional tunnels and slow exchange for within-tunnel ions. These exchange results provide corroboration for the geometrically determined site assignment in the 1-D spectrum, as well as support the notion of limited ion mobility in the Mn-phase resulting in poor electrochemical capability.

1. INTRODUCTION The development of sustainable energy alternatives is necessary in order to meet growing global energy demands. Lithium ion batteries have achieved considerable success in the portable electronics industry as a result of their ability to provide relatively nontoxic, high energy density, renewable power.1 Despite this success, the existing infrastructure must be improved in order for this technology to be a truly viable alternative to traditional energy sources such as fossil fuels. The cathode plays a critical role in determining battery performance, and as a result the optimization of this component is of particular importance. Iron phosphate structure derivatives were initially popularized by Goodenough and co-workers in the 1990s as a structurally robust polyanionic alternative to the layered oxides for Li cathode frameworks.2,3 To date, LiFePO4 (LFP) remains the most promising in the phosphate class as it exhibits high capacity and electrochemical reversibility in addition to being environmentally appealing.2 It has thus been successful despite suffering from inherently low electronic conductivity,4 and has motivated the search for other polyanionic materials that might exhibit performance comparable to that of LFP. The class of polyanionic materials has © XXXX American Chemical Society

been extended beyond simple phosphate structures to include pyrophosphates (Li 2 MP 2 O 7 ), 5 − 7 fluorophosphates (Li2MPO4F),8−11 and fluorosulfates (LiMSO4F),12−14 where in all cases the Fe-containing substituent is often desirable due to the low associated cost and low environmental impact relative to other metals. The pyrophosphate family of materials of the type Li2MP2O7, where M = Fe, Mn, is an increasingly interesting prospect as they have been shown to demonstrate a high operating voltage, as well as the ability to house two charge carrying Li ions per transition metal center, thus increasing theoretical capacities.5−7,15−18 Zhou et al.7 have reported highly reversible electrochemical cycling of the Fe-based pyrophosphate, with the ability to reproducibly remove a single lithium ion from the structure to generate the metastable LiFeP2O7 phase. However, despite promisingly high redox potentials, the Mn-containing phase appeared to exhibit consistently poor cycling behavior. This lack of electrochemical ability in Li2MnP2O7 is in Received: April 30, 2015 Revised: June 25, 2015

A

DOI: 10.1021/acs.jpcc.5b04173 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

One-dimensional SI has been advantageous for studying other Li materials exhibiting fast relaxation such as Li2VPO4F,23 LiVOPO4,31 and Li3Fe2(PO4)3,25 where the latter in particular demonstrated relatively fast ion dynamics relative to nuclear relaxation times. Li2MnP2O7 suffers both from short spin lifetimes and slow Li ion exchange rates, making the separation of both phenomena particularly challenging. In addition to short T1’s, the lack of peak separation makes the use of 2D EXSY difficult even for qualitative analysis, as the low signal-tonoise at mixing times sufficient for exchange is coupled with poor site resolution. The use of SI was therefore essential, as it allowed for the retention of site-specific information that lead to a reasonably quantitative determination of rate constants. The SI experiment consists of an initial shaped selective inversion shaped pulse at the site of interest, following which the inverted spins are allowed to exchange with noninverted spins, with a subsequent 90° readout pulse. It allows for the rapid collection of site-specific data points at many mixing times, thus facilitating a fit with an exchange matrix that accurately describes all rate processes in the system. The SI experiment utilized, although not inherently immune to effects due to relaxation, enables the collection of data at a variety of mixing times without the signal intensity loss suffered by 2D EXSY. This data can ultimately be modeled as a function of both the T1 relaxation rate as well as the unique ion-hopping rate for each Li exchange pair. The inclusion of all possible rate processes (both relaxation and chemical exchange between mobile sites) allows for a more quantitative treatment of the data.32,33 Although the aforementioned challenges make it difficult to obtain rate constants with small standard deviations, a representation of the exchange pathways can nevertheless be confidently attained. On account of the considerable broadening of the 6Li resonances for all Fe-containing phases in this series, exchange information was not obtained for these materials. However, due to the structural similarities between the Mn and Fe analogues, similar ion diffusion pathways are expected for all members in the mixed-metal series, thus allowing Li2MnP2O7 to act as a representative structure for this class of materials.

agreement with observations for other Mn phosphates, in particular LiMnPO4, where electrochemical performance pales in comparison to its olivine LiFePO4 counterpart.19,20 The mechanism for poor cyclability in the Mn pyrophosphate analog is as of yet unknown, and has been hypothesized to result from either a structural distortion during charging, or high kinetic barriers.6,7 Tamaru et al.6 have demonstrated that the Mn phase does in fact exhibit redox activity at 4.45 V with slow cycling rates, albeit with limited capacity, suggesting a possible kinetic barrier for Li ion diffusion through the cathode material. The encouragingly high redox potential of the entire Li2Mn1−yFeyP2O7 class of materials prompts the elucidation of their ion diffusion pathways as this would provide valuable insight into structural properties that contribute to electrochemical performance. Solid-state NMR (ssNMR) is an invaluable tool for the study of such materials, particularly when probing their ion mobility and ion diffusion pathways. The ability to measure site-specific dynamics with methods such as two-dimensional exchange spectroscopy (2D EXSY)21,22 and one-dimensional selective inversion (SI)23−25 makes this an excellent method for describing electrochemical performance as a result of Li ion kinetics. Although the paramagnetic nature of many of these materials traditionally makes them particularly challenging to study by ssNMR, the influence of unpaired electron spin density at the ion of interest actually allows for a more thorough structural characterization in some cases as the paramagnetism dramatically increases the Li chemical shift range, often allowing for resolution of unique crystallographic Li environments by NMR. The dominance of the paramagnetic effect on the observed Li shifts in these materials permits the interpretation of the spectra on the basis of unpaired electron spin density transfer to the nucleus of interest. The most rigorous approach involves the use of DFT calculations in an effort to determine electronic contributions to the nucleus of interest and assign the observed chemical shifts to crystallographic environments in this way.26−28 In the absence of DFT, these largely paramagnetically shifted sites can be assigned with sufficient knowledge of their crystallographic arrangement. Using the Goodenough−Kanamori rules, Grey and co-workers found empirical evidence that lithium−oxygen-metal bond angles are closely related to the observed paramagnetic shift.29,30 Two electron spin density transfer mechanisms are generally observed, polarization and delocalization, where contributions from the latter are found to give rise to positive paramagnetic shifts.29 While DFT is often preferred in the phosphates due to their lack of exact 90° or 180° contacts, there has been some success in assigning chemical shifts in similar materials using geometric arguments.26−28 The reported bond angles and Li−O distances are used here to identify these spin transfer pathways in Li2MnP2O7 and to assign peaks in the NMR spectrum to crystallographic Li environments, results that will be ultimately validated by the identified ion exchange pathways. Not only is the observed shift dominated by the interaction with unpaired e− spin density, but nearby nuclei often possess uncharacteristically short spin−lattice relaxation (T1) rates, an effect that can prove problematic for exchange studies that require the spin lifetimes to exceed the time scale of the exchange process. In the event that T1’s are on a time scale comparable to that of the exchange, careful separation of relaxation from exchange information is necessary for an accurate measure of ion mobility.

2. EXPERIMENTAL SECTION 2.1. Sample Preparation. 6Li enriched Li2Mn1−yFeyP2O7 samples of varying composition (y = 0, 0.2, 0.5, 0.8, 1) were prepared by the “wet” method described by Zhou et al.7 Stoichiometric amounts of 6Li(CH3CO2), Fe(CH3CO2)2, Mn(CH3CO2)2, and NH4H2PO4 were stirred in aqueous solution (0.02 M Li) and heated until dry. The sample was placed in an evacuated drying oven at 90 °C for 12 h. Using a tube furnace with a constant flow of 5% H2/N2 at 5 psi, the sample was heated to 600 °C for 30 h. The sample was subsequently reground and heated to 600 °C for 12 h. 6 LiFeP2O7 was prepared by the previously reported high temperature synthesis as described by Padhi et al.2 Stoichiometric amounts of Fe2O3, NH4H2PO4, and 6Li2CO3 were ground and heated at 200 to 300 °C. The sample was then reground and heated in a tube furnace to 850 °C for 24 h. 2.2. Solid-State NMR. All 6Li MAS NMR spectra were acquired at a Larmour frequency of 44.1 MHz on a Bruker AV300 spectrometer using 1.8 mm diameter rotors in a doubleresonance probe. A Hahn-echo pulse sequence with a 90° pulse of 3.5 μs and a recycle delay of 200 ms was used to acquire the 6 Li MAS spectra. 6Li selective inversion recovery spectra were acquired using a 180°−τmix−90° pulse sequence, where the B

DOI: 10.1021/acs.jpcc.5b04173 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C 180° Gaussian shaped pulse was 1 ms in duration with a soft pulse power of 0.109 W. The mixing time was varied across a series of experiments and was chosen to range from 5 μs to 200 ms. All variable temperature 6Li MAS NMR spectra were acquired at a Larmor frequency of 44.1 MHz on a Bruker AV300 spectrometer using a custom built double-resonance probe supporting rotors 1.8 mm in diameter. Temperatures were calibrated using Sm2Sn2O7 according to the method discussed previously by Grey et al.34 2.3. Data Analysis with CIFIT. Peak areas acquired as a function of mixing time by the SI method were analyzed by the CIFIT program developed by Bain and Cramer.33 CIFIT utilizes a rate matrix describing the relaxation properties of each spin under the influence of chemical exchange, the theory of which is discussed in detail elsewhere.32,35 The program varies relevant parameters using the Levenberg−Marquardt algorithm and compares with experimental data attempting to minimize the sum of the squares of the difference.

Figure 2. 6Li MAS (25 kHz) spectrum of Li2MnP2O7 at 298 K and the deconvoluted spectrum indicating the four partially resolved Li resonances corresponding to the four unique crystallographic sites in the structure.

the Goodenough−Kanamori rules, as previously employed for the assignment of Li resonances in other paramagnetic compounds.29 The transfer of electron spin density from the half-filled t2g orbital of paramagnetic Mn(II) to the empty Li 2s orbital via a delocalization mechanism causes a shift in the Li resonance where the efficiency of this transfer is maximized when the Li−O−Mn orbitals overlap at angles close to 90 or 180°.30 While monoclinic Li2MnP2O7 contains no Li−O−Mn angles close to 180°, there are angles close to 90°, where this 90° geometric arrangement is expected to give rise to an increase in the observed isotropic shift.29,30 The assignment of the observed NMR signals to crystallographic sites are predicted herein, utilizing previously reported Li−O−Mn bond angles and Li−O bond lengths.5 Therefore, Li3, having two Li−O−Mn angles approaching 90° (89.53 and 94.85°) and the shortest Li−Mn distances (3.01 Å), is assigned to the highest frequency site at 44 ppm and labeled site D in the 6Li NMR spectrum. In contrast Li2, having no Li−O−Mn bond angles close to 90°, is ascribed to the lowest frequency resonance at 2.5 ppm (denoted site A). The Li−O−Mn contacts for each lithium site are depicted in Figure 3, with the relevant bond angles, bond lengths, and peak assignments for

3. RESULTS AND DISCUSSION Li2MnP2O7 was first synthesized in 2008 by Adam et al.5 where the 3D pyrophosphate framework reportedly consisted of MnO5 and MnO6 polyhedra interconnected through P2O7 groups, generating a tunnel-like structure that houses the four crystallographically unique Li ions, with Li1 and Li2 sharing one tunnel and Li3 and Li4 occupying another. The structure is depicted in Figure 1 and illustrated in the VESTA visualization

Figure 1. Unit cell of monoclinic Li2MnP2O7 along the a-axis. MnO5 trigonal bipyramids and MnO6 octahedra (purple) are connected to PO4 tetrahedra (gray). Li1, Li2, Li3, and Li4 sit in tunnels formed by the MnO5, MnO6, and PO4 units.6

program.36 The powder X-ray diffraction pattern of 6-Li enriched Li2MnP2O7 (provided in the Supporting Information) agrees with that of 7Li2MnP2O7 reported by Adam et al.,6 thus confirming the successful preparation of the isotopically enriched pyrophosphates via the “wet method” described by Zhou et al.7 Although the structure is well-known by diffraction, detailed information about local environments and dynamics of particular atoms is lacking. Solid-state NMR is used here to probe local environments in a series of Li2Mn1−yFeyP2O7 materials. 3.1. Characterization of Mixed Metal Pyrophosphates. The 6Li MAS NMR spectrum of Li2MnP2O7 is shown in Figure 2, where the four Li resonances corresponding to the distinct crystallographic positions in Li2MnP2O7 are partially resolved. The four resonances (at 44, 32, 7, and 2.5 ppm) have been assigned to their respective crystallographic positions based on

Figure 3. Local environments for the four Li sites in Li2MnP2O7, showing Li−O bonds and Li−O−Mn contacts. Li1 (a) and Li4 (d) show distorted square pyramidal geometry, Li2 (b) shows nearly tetrahedral geometry, and Li3 (c) shows distorted tetrahedral geometry. Only Li−O−Mn contacts with Li−Mn distances less than 3.92 Å are shown. C

DOI: 10.1021/acs.jpcc.5b04173 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Although the four Li sites are partially resolved in the Li2MnP2O7 spectrum, as iron content increases the signal broadens and the resonances overlap, removing the site resolution. The measured spin−lattice (T1) and spin−spin (T2) nuclear relaxations of the pyrophosphates are given in Table 3. The T1s were found to increase with iron content

the four crystallographic Li sites outlined in Table 1. A deconvolution of the peaks in the 6Li MAS spectrum of Table 1. Summary of Crystallographic Data for Li2MnP2O7a Li site Li1B

Li2A Li3D

Li4C

a

Li−O−Mn angle (deg) 99.37 88.20 94.23 96.10 110.74 105.67 89.53 94.85 111.64 100.37 90.86 116.77 119.26

(Li1−O6−Mn2) (Li1−O13−Mn2) (Li1−O13−Mn1) (Li1−O14−Mn1) (Li1−O3−Mn1) (Li2−O14−Mn1) (Li3−O12−Mn2) (Li3−O8−Mn2) (Li3−O1−Mn1) (Li4−O4−Mn2) (Li4−O8−Mn2) (Li4−O1−Mn1) (Li4−O12−Mn2)

distance to Mn (Å)

Li−O distance (Å)

3.171 3.171 3.352 3.352 3.463 3.309 3.010 3.010 3.697 3.175 3.175 3.548 3.915

1.982(3) 2.328(4) 2.328(4) 2.351(4) 2.092(4) 1.999(4) 1.951(6) 1.966(6) 2.288(5) 1.970(5) 2.333(4) 1.984(5) 2.229(4)

Table 3. Spin−Lattice (T1) and Spin−Spin (T2) Relaxation Times for the Pyrophosphates Li2Mn1‑yFeyP2O7 y y y y y

Li 2 MnP 2 O 7 enables the determination of relative site occupancies. Integration over the entire spinning sideband manifold reveals an equal population of the two sites in each of the two tunnels, although the integrated intensities of Li3 and Li4 are 84% that of Li1 and Li2 (Table 2). The set of 1D 6Li Table 2. Peak Intensities of the Four Lithium Resonances in Li2MnP2O7

3D 4C 1B 2A

integral over entire spinning sideband manifold

intensity (normalized to site 1B)

× × × ×

0.84 0.84 1.02 1.00

1.12 1.13 1.37 1.34

1011 1011 1011 1011

0 0.2 0.5 0.8 1

avg T1 (ms) 3.06 4.33 4.70 10.97 17.26

± ± ± ± ±

0.43 0.03 0.05 0.07 0.06

avg T2 (ms) 1.08 0.69 2.04 2.99 1.82

± ± ± ± ±

0.15 0.06 0.26 0.48 0.19

(3.06 ± 0.43 ms for Li2MnP2O7 and 17.26 ± 0.06 ms for Li2FeP2O7) with the differences in T2s being negligible. As such, the differences in line width cannot be attributed to a difference in nuclear relaxation properties across samples. Broadening of the 6Li resonance can also be caused by the nuclear spin interaction with the time averaged magnetic moments of the electronic spins at the paramagnetic transition metals. These hyperfine interactions are known to occur via either a through-space (dipolar) or through-bond (Fermicontact) mechanism.30 When the magnetic moment of the electronic spins is anisotropic there is an additional dipolar interaction that causes inhomogeneous line broadening. The symmetric electron configuration of Mn2+ (t2g3eg2) gives rise to an isotropic magnetic moment. In contrast, the asymmetric electron configuration of Fe2+ (t2g4eg2) results in an anisotropic magnetic moment, thereby causing additional line broadening in the iron(II) containing pyrophosphates. The effect of this anisotropy of the magnetic moment on line broadening is also demonstrated by simply comparing the 6Li MAS spectrum of LiFeP2O7 to that of Li2FeP2O7 (Figure 5). Akin to the Mn2+ example, the electronic configuration of Fe3+ (t2g3eg2) in

Li−O−Mn contacts with Li−Mn distances less than 3.92 Å.5

site

= = = = =

MAS NMR spectra of the pyrophosphate series (Li2Mn1−yFeyP2O7, 0 ≤ y ≤ 1) are shown in Figure 4.

Figure 4. 6Li MAS NMR spectra of Li2Mn1−yFeyP2O7 (y = 0, 0.2, 0.5, 0.8, 1) with MAS = 25 kHz. Full spectra and isotropic peaks are shown in (b) and (a), respectively. The asterisk (*) indicates an unidentified impurity in the Li2MnP2O7 material.

Figure 5. 6Li MAS NMR spectra of singly lithiated LiFeP2O7 (blue) and doubly lithiated Li2FeP2O7 (black) with MAS = 25 kHz. Full spectra and isotropic peaks are shown in (b) and (a), respectively. D

DOI: 10.1021/acs.jpcc.5b04173 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 6. (a) One-dimensional 6Li spectrum of Li2MnP2O7 with shaded area representing the bandwidth of the Gaussian inversion pulse. (b) Stack plot of 6Li MAS spectra collected using the SI experiment with variable mixing times.

demonstrated here, the inability to qualitatively observe this transient attenuation of signal does not hinder the ability of the experiment to capture ion hopping rates and thus overall ion mobility. The SI experiment was carried out with Li2MnP2O7 at room temperature and 350 K. When sites are in exchange with one another, the perturbation of spins at one site via an inversion pulse will be observed as an attenuation of intensity at exchanging sites with appropriate mixing times in the SI experiment. Therefore, in addition to an intensity buildup of the inverted resonance, a typical “transient well” curve will be obtained for noninverted sites as a result of their chemical exchange with the inverted site. As is depicted in Figure 7, this

LiFeP2O7 is symmetric, corresponding to an isotropic magnetic moment, and in consequence the NMR spectrum of LiFeP2O7 is well resolved. 3.2. Selective Inversion and CIFIT Results for Li2MnP2O7. The use of SI NMR as a method for characterizing ion dynamics in Li ion cathode materials has been extremely useful in determination of multisite exchange pathways.23,24,37 Traditional 2D exchange spectroscopy has proven difficult when relaxation times are very fast, and a lack of sufficient resolution can further muddle 2D EXSY spectra. As such, SI was employed as a method for obtaining rate constants for exchange between the four crystallographically unique sites in the Mn pyrophosphate material. The lack of optimal site resolution in the 1D MAS spectrum of Li2MnP2O7 makes the successful application of a purely selective pulse near impossible, thus despite attempting to invert only a single resonance the nearby peak(s) will invariably be affected by the inversion pulse, as depicted in Figure 6. This is in contrast to work with other phosphates that have enjoyed sufficient siteresolution to achieve near-perfect selectivity.23,25 Fortunately, this can be accounted for in the relaxation matrix used to fit the experimental data by including the initial attenuated intensity of the affected peaks in the fit. In fact, upon testing various selective pulse lengths and shapes, the extent to which nearby sites were affected by the selective pulse had little effect on the resulting rate constant obtained from the fit. Two examples are provided in Table 4 where, despite a difference in inversion Table 4. Effect of Inversion Efficiency on Obtained Rate Constant and Goodness of Fit as Measured by a χ2 Value.a inversion percentage of site D 41% 63%

rate constant, kCD (s−1)

χ2 (from CIFIT)

35 ± 25 50 ± 25

0.023 0.029

a

Two experiments whereby site LiD is inverted to 41% and 61% gave rate constants identical within error, without a significant change in the fit to experimental data.

efficiency, rate constants are found to be equal within error. Ideally, a plot of the intensity as a function of mixing time should give a series of “build-up” and “transient well” curves for the inverted and noninverted sites, respectively. This transient decrease in intensity arises from the exchange between the inverted and noninverted sites, and the fit of this portion of the curve yields a rate constant for this exchange process. As will be

Figure 7. Normalized peak area as a function of mixing time at (a) 298 K and (b) 350 K upon inversion of site D. Solid lines demonstrate the CIFIT results. E

DOI: 10.1021/acs.jpcc.5b04173 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

NMR experiments to identify ion diffusion pathways in this material. It was ascertained that slow ion hopping occurs between Li ions residing within the same tunnel, but not between tunnels. The relatively slow ion exchange between only the closest Li ion pairs is consistent with the poor electrochemical performance exhibited by the Mn-containing analog, particularly at fast cycling rates. This result provides important insight into the Li ion mobility pathways in the pyrophosphate class of materials, an observation that could be extended to other, similar structural motifs.

transient well is not readily observed for the red and blue curves (corresponding to the noninverted sites in this example) contrary to expectations. This is attributed to the combination of short nuclear relaxation times with relatively slow exchange rates for the mobile Li sites. Despite a lack of defined transient well behavior for noninverted spins in the pyrophosphate system, CIFIT can still effectively model the intensity as a function of mixing time using rate constants for the appropriate exchange processes in combination with predetermined nuclear relaxation rates. At all available temperatures, resonances LiA and LiB remained too significantly overlapped to obtain any Li ion hopping rates between these sites. However, the rate constant for LiD-LiC exchange was found to sit between 45−100 Hz over the investigated temperature range. Initially allowing all rate constants to be varied by the CIFIT fitting procedure yielded results with unreasonably high error bars (>100%) and even negative rate constants in some instances. This prompted a reevaluation of the procedure in which all parameters included in the rate matrix were allowed to vary, and indeed by limiting the available exchange partners more reasonable fits and error bars were obtained. The exchange pairs were therefore restricted in the rate matrix to fit only LiC-LiD exchange, as fits for other exchange pairs nearly always gave rise to a negligible and/or negative rate constant from CIFIT and were thus assumed to be zero. The exchange rates allow us to corroborate the aforementioned site assignment in this material, where Li resonances were initially assigned to crystallographic positions by analysis of Li−O−Mn contacts. It is worth acknowledging that LiA-LiB exchange is not quantified herein due to a lack of spectral resolution, however, as a result of their geometric similarity to the LiC−LiD pair, they are expected to exchange with each other on a similar time scale. On account of the exchange being exclusively observed between two pairs of Li ions, it can be presumed that the exchangeable Li ions reside within the same two-dimensional tunnel rather than in opposing tunnels. As Li ions sharing a tunnel are ∼3 Å apart, compared to >5 Å for cross-tunnel distances, it would be unlikely to observe exchange between Li ions residing in different tunnels without any within-tunnel ion hopping. We can therefore use the selective inversion results to support the initial peak assignments, where we can ascribe peaks A and B to Li1 and Li2 respectively and similarly, C and D to Li3 and Li4. If, as the results indicate, LiC and LiD reside in a tunnel distinct from LiA or LiB, the exchange data suggests that Li ion mobility occurs only within tunnels in the Li2MnP2O7 structure, and not between them. While the combination of slow Li ion hopping and fast nuclear relaxation makes exchange studies by NMR challenging, we demonstrate here that reasonable kinetics information can still be obtained to provide some insight into Li ion diffusion pathways in the monoclinic pyrophosphate structures.

■

ASSOCIATED CONTENT

S Supporting Information *

X-ray powder diffraction data on the pyrophosphate materials. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.5b04173.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: 905-525-9140 x24176. Fax: 905-522-2509. Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Tarascon, J.-M.; Armand, M. Issues and Challenges Facing Rechargeable Lithium Batteries. Nature 2001, 414, 359−367. (2) Padhi, A. K.; Nanjundaswamy, K. S.; Goodenough, J. B. PhosphoOlivines as Positive-Electrode Materials for Rechargeable Lithium Batteries. J. Electrochem. Soc. 1997, 144, 1188−1194. (3) Padhi, A. K.; Nanjundaswamy, K. S.; Masquelier, C.; Masque; Okada, S.; Goodenough, J. B. Effect of Structure on the Fe3+/Fe2+ Redox Couple in Iron Phosphates. J. Electrochem. Soc. 1997, 144, 1609−1613. (4) Ait Salah, A.; Mauger, A.; Zaghib, K.; Goodenough, J. B.; Ravet, N.; Gauthier, M.; Gendron, F.; Julien, C. M. Reduction Fe3+ of Impurities in LiFePO4 From Pyrolysis of Organic Precursor Used for Carbon Deposition. J. Electrochem. Soc. 2006, 153, A1692−A1701. (5) Adam, L.; Guesdon, A.; Raveau, B. A New Lithium Manganese Phosphate with an Original Tunnel Structure in the A2MP2O7 Family. J. Solid State Chem. 2008, 181, 3110−3115. (6) Tamaru, M.; Barpanda, P.; Yamada, Y.; Nishimura, S.-I.; Yamada, A. Observation of the Highest Mn3+/Mn2+ Redox Potential of 4.45 v in a Li2MnP2O7 Pyrophosphate Cathode. J. Mater. Chem. 2012, 22, 24526−24529. (7) Zhou, H.; Upreti, S.; Chernova, N. A.; Hautier, G.; Ceder, G.; Whittingham, M. S. Iron and Manganese Pyrophosphates as Cathodes for Lithium-Ion Batteries. Chem. Mater. 2011, 23, 293−300. (8) Ramesh, T. N.; Lee, K. T.; Ellis, B. L.; Nazar, L. F. Tavorite Lithium Iron Fluorophosphate Cathode Materials: Phase Transition and Electrochemistry of LiFePO4F−Li2FePO4F. Electrochem. SolidState Lett. 2010, 13, A43−A47. (9) Ellis, B. L.; Makahnouk, W. R. M.; Rowan-Weetaluktuk, W. N.; Ryan, D. H.; Nazar, L. F. Crystal Structure and Electrochemical Properties of A2MPO4F Fluorophosphates (a = Na, Li; M = Fe, Mn, Co, Ni). Chem. Mater. 2010, 22, 1059−1070. (10) Kim, S.-W.; Seo, D.-H.; Kim, H.; Park, K.-Y.; Kang, K. A Comparative Study on Na2MnPO4F and Li2MnPO4F for Rechargeable Battery Cathodes. Phys. Chem. Chem. Phys. 2012, 14, 3299−3303. (11) Lee, S.; Park, S. S. Lithium Transition Metal Fluorophosphates (Li2CoPO4F and Li2NiPO4F) as Cathode Materials for Lithium Ion Battery From Atomistic Simulation. J. Solid State Chem. 2013, 204, 329−334. (12) Barpanda, P.; Ati, M.; Melot, B. C.; Rousse, G.; Chotard, J.-N.; Doublet, M.-L.; Sougrati, M. T.; Corr, S. A.; Jumas, J.-C.; Tarascon, J.M. A 3.90 v Iron-Based Fluorosulphate Material for Lithium-Ion

4. CONCLUSIONS The series of mixed-metal pyrophosphates of the form Li2MnyFe1−yP2O7 were investigated by ssNMR spectroscopy in an effort to understand ion dynamic effects on electrochemical performance in this class of materials. 6Li ssNMR spectra of the mixed-metal species are reflective of a distinct change in the electronic structure of the transition metal, where even relatively low iron content resulted in significantly broadened 6Li resonances, making site resolution in the Fecontaining phases unfeasible. In contrast, the unique Li environments in the pristine Mn phase can be analyzed by SI F

DOI: 10.1021/acs.jpcc.5b04173 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C Batteries Crystallizing in the Triplite Structure. Nat. Mater. 2011, 10, 772−779. (13) Recham, N.; Chotard, J.-N.; Dupont, L.; Delacourt, C.; Walker, W.; Armand, M.; Tarascon, J.-M. A 3.6 v Lithium-Based Fluorosulphate Insertion Positive Electrode for Lithium-Ion Batteries. Nat. Mater. 2010, 9, 68−74. (14) Tripathi, R.; Ramesh, T. N.; Ellis, B. L.; Nazar, L. F. Scalable Synthesis of Tavorite LiFeSO4F and NaFeSO4F Cathode Materials. Angew. Chem., Int. Ed. 2010, 49, 8738−8742. (15) Furuta, N.; Nishimura, S.-I.; Barpanda, P.; Yamada, A. Fe 3+/Fe 2+Redox Couple Approaching 4 v in Li2− X(Fe1−yMny)P2O7 Pyrophosphate Cathodes. Chem. Mater. 2012, 24, 1055−1061. (16) Shimizu, D.; Nishimura, S.-I.; Barpanda, P.; Yamada, A. Electrochemical Redox Mechanism in 3.5 v Li 2- xFeP 2O 7(0 ≤ X≤ 1) Pyrophosphate Cathode. Chem. Mater. 2012, 24, 2598−2603. (17) Barpanda, P.; Rousse, G.; Ye, T.; Ling, C. D.; Mohamed, Z.; Klein, Y.; Yamada, A. Neutron Diffraction Study of the Li-Ion Battery Cathode Li2FeP2O7. Inorg. Chem. 2013, 52, 3334−3341. (18) Tamaru, M.; Chung, S. C.; Shimizu, D.; Nishimura, S.-I.; Yamada, A. Pyrophosphate Chemistry Toward Safe Rechargeable Batteries. Chem. Mater. 2013, 25, 2538−2543. (19) Delacourt, C.; Laffont, L.; Bouchet, R.; Wurm, C.; Leriche, J.-B.; Morcrette, M.; Tarascon, J.-M.; Masquelier, C. Toward Understanding of Electrical Limitations (Electronic, Ionic) in LiMPO4 (M=Fe, Mn) Electrode Materials. J. Electrochem. Soc. 2005, 152, A913−A919. (20) Li, G.; Azuma, H.; Tohda, M. LiMnPO4 as the Cathode for Lithium Batteries. Electrochem. Solid-State Lett. 2002, 5, A135−A137. (21) Cahill, L. S.; Chapman, R. P.; Britten, J. F.; Goward, G. R. 7Li NMR and Two-Dimensional Exchange Study of Lithium Dynamics in Monoclinic Li3V2(PO4)3. J. Phys. Chem. B 2006, 110, 7171−7177. (22) Jeener, J.; Meier, B. H.; Bachmann, P.; Ernst, R. R. Investigation of Exchange Processes by Two-Dimensional NMR Spectroscopy. J. Chem. Phys. 1979, 71, 4546−4553. (23) Davis, L. J. M.; Ellis, B. L.; Ramesh, T. N.; Nazar, L. F.; Bain, A. D.; Goward, G. R. 6Li 1D EXSY NMR Spectroscopy: a New Tool for Studying Lithium Dynamics in Paramagnetic Materials Applied to Monoclinic Li2VPO4F. J. Phys. Chem. C 2011, 115, 22603−22608. (24) Davis, L. J. M.; Goward, G. R. Differentiating Lithium Ion Hopping Rates in Vanadium Phosphate Versus Vanadium Fluorophosphate Structures Using 1D 6Li Selective Inversion NMR. J. Phys. Chem. C 2013, 117, 7981−7992. (25) Smiley, D. L.; Davis, L. J. M.; Goward, G. R. An Improved Understanding of Li+ Hopping Pathways and Rates in Li3Fe2(PO4)3 Using Selective Inversion 6Li NMR Spectroscopy. J. Phys. Chem. C 2013, 117, 24181−24188. (26) Castets, A.; Carlier, D.; Trad, K.; Delmas, C.; Ménétrier, M. Analysis of the 7Li NMR Signals in the Monoclinic Li3Fe2(PO4)3 and Li3V2(PO4)3 Phases. J. Phys. Chem. C 2010, 114, 19141−19150. (27) Kim, J.; Middlemiss, D. S.; Chernova, N. A.; Zhu, B. Y. X.; Masquelier, C.; Grey, C. P. Linking Local Environments and Hyperfine Shifts: a Combined Experimental and Theoretical 31P and 7Li Solid-State NMR Study of Paramagnetic Fe(III) Phosphates. J. Am. Chem. Soc. 2010, 132, 16825−16840. (28) Middlemiss, D. S.; Ilott, A. J.; Clément, R. J.; Strobridge, F. C.; Grey, C. P. Density Functional Theory-Based Bond Pathway Decompositions of Hyperfine Shifts: Equipping Solid-State NMR to Characterize Atomic Environments in Paramagnetic Materials. Chem. Mater. 2013, 25, 1723−1734. (29) Grey, C. P.; Lee, Y. J. Lithium MAS NMR Studies of Cathode Materials for Lithium-Ion Batteries. Solid State Sci. 2003, 5, 883−894. (30) Grey, C. P.; Dupré, N. NMR Studies of Cathode Materials for Lithium-Ion Rechargeable Batteries. Chem. Rev. 2004, 104, 4493− 4512. (31) Davis, L. J. M.; He, X. J.; Bain, A. D.; Goward, G. R. Solid State Nuclear Magnetic Resonance. Solid State Nucl. Magn. Reson. 2012, 42, 26−32. (32) Bain, A. D.; Cramer, J. A. A Method for Optimizing the Study of Slow Chemical-Exchange by Nmr Spin-Relaxation Measurements -

Application to Tripodal Carbonyl Rotation in a Metal-Complex. J. Magn. Reson., Ser. A 1993, 103, 217−222. (33) Bain, A. D.; Cramer, J. A. Slow Chemical Exchange in an EightCoordinated Bicentered Ruthenium Complex Studied by OneDimensional Methods. Data Fitting and Error Analysis. J. Magn. Reson., Ser. A 1996, 118, 21−27. (34) Grey, C. P.; Cheetham, A. K.; Dobson, C. M. TemperatureDependent Solid-State 119Sn- MAS NMR of Nd2Sn2O7, Sm2Sn2O7, and Y1.8Sm0.2Sn2O7. Three Sensitive Chemical-Shift Thermometers. J. Magn. Reson., Ser. A 1993, 101, 299−306. (35) Bain, A. D.; Fletcher, D. A. Selective-Inversion Experiments Applied to Chemical Exchange in Coupled Spin Systems. Mol. Phys. 1998, 95, 1091−1098. (36) Momma, K.; Izumi, F. VESTA 3for Three-Dimensional Visualization of Crystal, Volumetric and Morphology Data. J. Appl. Crystallogr. 2011, 44, 1272−1276. (37) Smiley, D. L.; Davis, L. J. M.; Goward, G. R. An Improved Understanding of Li+ Hopping Pathways and Rates in Li3Fe2(PO4)3 Using Selective Inversion 6Li NMR Spectroscopy. J. Phys. Chem. C 2013, 117, 24181−24188.

G

DOI: 10.1021/acs.jpcc.5b04173 J. Phys. Chem. C XXXX, XXX, XXX−XXX