Christopher
Sole
,
Nicholas E.
Drewett
and
Laurence J.
Hardwick
*
Stephenson Institute for Renewable Energy, Department of Chemistry, University of Liverpool, L69 7ZF, UK. E-mail: laurence.hardwick@liv.ac.uk
First published on 8th May 2014
The first and second order Raman spectra of graphite during the first lithiation and delithiation have been investigated in a typical lithium-ion battery electrolyte. In situ, real-time Raman measurements under potential control enable the probing of the graphitic negative electrode surface region during ion insertion and extraction. The experimental results reveal the staging formation of a single particle within a free standing graphitic electrode. In particular, the in situ behaviour of the double resonance 2D band during the lithiation and delithiation of graphitic carbon has not been previously reported. The 2D band was observed to shift from 2681 to 2611 cm−1 and the band shape transformed into a single Lorentzian from 0.24 to 0.15 V vs. Li/Li+, providing further information on the electronic structure and C–C bonding of stage 3 and 4 graphite intercalation compounds. The behaviour of the 2D band is in keeping with the Daumas–Hérold model of electrochemically derived intercalation, where the graphene layers are flexible and deform around domains of intercalating lithium ions.
Investigations into electrode reactions are considerably improved when electrode/solution interface spectra can be recorded as the electrochemical response is obtained. In situ Raman spectroscopy is a convenient and non-destructive tool for studying lithiation/delithiation processes within numerous battery electrodes, and it has been utilised to characterise a number of insertion materials over the past couple of decades, as summarised by two recent review articles.5,6
Highly crystalline graphitic materials are routinely used as the negative electrode in lithium-ion batteries. Their positive features include a high, reversible specific charge of up to the theoretical value of 372 A h kg−1 (of carbon) for the formation of the donor graphite intercalation compound (GIC) LiC6, a good cycling stability (for portable electronic applications), a high electronic conductivity and low cost. However, problems persist with insufficient rate performance and electrode degradation over time. In addition, due to the low potential of intercalation (<0.2 V vs. Li/Li+), lithium plating is a major safety risk at high charge rates.
The staging process of lithium intercalation into carbon was observed in a typical Li-ion battery electrolyte by Inaba et al.7 during their pioneering in situ Raman microscopy studies. They observed that lithium intercalation proceeds through a series of staged graphite intercalation compounds (GICs), classified by a stage index, n, which represents the number of graphene layers separating the layers of intercalated ions. The Raman spectra for GICs with stage n > 2 are known to exhibit a doublet G band. The lower (E2g2(i)) and upper (E2g2(b)) frequency components are correspondingly associated with carbon-atom vibrations in interior graphite layers (not adjacent to the intercalate layer planes) and in bounding graphite layers (adjacent to the intercalate planes). The split in the E2g2 mode upon intercalation occurs primarily from changes in symmetry at the boundary layer, and secondarily from the electronic effects of the intercalate molecule. The E2g2(i) band disappears for stage 1 and 2, where no graphite interior layer exists.
A quantitative measure of the intercalation stage index, n, can be derived from the relative intensities of the Raman doublet, R, by the following equation:8
(1) |
The staging process during lithium insertion into graphitic carbons had been previously proposed by Dahn et al.9 following detailed in situ powder X-ray diffraction (PXRD) studies. However, whilst PXRD probes the bulk electrode, Raman microscopy offers the ability to follow the processes of lithiation in individual carbon particles, or different areas of a single particle within the electrode.10–12
Improved understanding of electrochemical processes in functioning battery electrodes will require an insight into more localised lithiation mechanisms. Thus, a combination of surface and bulk characterisation is critical in realising the function of these materials under operating battery conditions.
Despite numerous in situ Raman studies of Li insertion into carbon,7,10–20 the effect on the 2D band has yet to be reported. The 2D band is the overtone of the D band. Whilst the D band requires the presence of defects for activation, the overtone originates from a process whereby momentum can always be conserved by two phonons with opposite wave vectors,21 hence the 2D band is always present. The 2D band (also referred to as G′ within the literature) has been used extensively in the research of sp2 carbons. It allows determination of the number and orientation of graphene layers in few layered graphene samples (1–5 layers),21 in addition to providing information on induced strain22–24 and charging/doping effects.25,26 In this study, we report on the behaviour of the 2D band during lithiation into microcrystalline graphite, in order to improve our understanding of both the model of Li+ intercalation and the electronic structure of GICs during lithium intercalation and extraction.
In order to acquire spectra with a good signal to noise ratio, the laser was focussed onto a suitable graphite particle as shown in Fig. 2(C), and the scattered light was collected from a volume of ca. 1–2 μm3. The point for in situ measurements is carefully designated. On certain areas of the graphite electrode there is minimal detection of Raman peaks from the electrolyte. Furthermore, to allow for the refocussing of the laser throughout the experiment, and checking that the selected measurement spot has been maintained, a particle or area of the electrode with distinguishable features must be selected.
Fig. 3 shows the first and second order Raman spectra for the pristine synthetic graphitic carbon measured in this study. The main peaks observed are the D, G and 2D bands, appearing at 1336 cm−1, 1580 cm−1 and ca. 2670 cm−1 respectively. The 1580 cm−1 peak (E2g2), called the G band after crystalline graphite, is the only easily accessible Raman active mode of the infinite lattice. There is a second Raman active mode E2g1 at (41 cm−1), but is difficult to observe due to its proximity to the Rayleigh line. The G mode is due to the relative motion of sp2 carbon atoms in rings as well as chains. The peak at 1336 cm−1 is named the D band, from disordered graphite, and can be attributed to the breathing motion of sp2 atoms in the rings at edge planes and defects in the graphene sheet. The origin of the D band has been discussed by Ferrari et al.21
Peak fitting of the 2D band gives two peaks in accordance with the literature28,29 that have been designated 2D(1) and 2D(2), at 2649 cm−1 and 2686 cm−1 respectively. It should be noted that Cançado et al.30 have proposed that the two-peak shape of the 2D band in bulk graphite actually arises from the convolution of an infinite number of peaks.
The load curve showing the first discharge and charge profile is shown in Fig. 4. As with other graphitic materials, the charge consumed for the formation of LiC6 in the first cycle exceeds the theoretical maximum of 372 A h kg−1. This is due to the partial reduction of electrolyte during the formation of a passivation film called the solid electrode interphase (SEI). The electrochemical performance of the in situ Raman cell matches the performance of similar graphitic materials previously reported in the literature.31
Fig. 4 A graph showing discharge and charge profiles for the microcrystalline graphite electrode in the in situ Raman cell against a metallic lithium counter/reference. Raman spectra acquisitions are marked with numbers corresponding to those shown in Fig. 5. |
The results from the first lithium intercalation and de-intercalation cycle are shown in Fig. 5. The first order in situ Raman spectra compare well to the reported literature.7,12,16 The potential (V) at which each spectrum was collected is displayed. All the spectra are base-line corrected and stacked arbitrarily up the y-axis to allow for clear visualisation. Bands due to the electrolyte are either not detected or negligible because of the employed confocal set-up, which allows a confocal resolution of 1–2 μm.3
Fig. 5 In situ Raman spectral series for the first lithium insertion and extraction into 6 μm graphitic carbon; the potential at each spectral acquisition is labelled to the right. |
At the open circuit potential (ca. 3.0 V vs. Li/Li+), three main bands are observed in the region between 1000–3000 cm−1: the G band at 1580 cm−1 and the 2D band at 2670 cm−1. A weak D band is noted at 1330 cm−1.
During lithiation, the series of first and second order band spectra may be divided into four stages of specific interest: the initial loss of D band intensity between potentials ca. 3.0–0.6 V; blue-shifting of the G band from 1580 cm−1 to 1590 cm−1, accompanied by the gradual weakening (and eventual disappearance) of the 2D(1) band intensity between ca. 0.6–0.2 V; splitting of the G band into the E2g2(i) (1575 cm−1) and E2g2(b) (1601 cm−1) bands ca. 0.20–0.15 V, along with a major red-shift of the 2D(2) band; and finally the appearance of a weak peak around 1370 cm−1 and the gradual loss of all distinct Raman peaks below ca. 0.10 V, associated with the formation of highly conductive, low stage number GICs.
The reverse spectral series (lithium de-intercalation) may also be split into four stages of specific interest: the weak band at 1370 cm−1 reappears from ca. 0.14 V along with a broad G band at 1592 cm−1; a weakening of the 1370 cm−1 band accompanies the growth, sharpening and blue-shift of the G band from 1592 to 1598 cm−1 to form the E2g2(b) band of the stage 2 GIC at ca. 0.15 V; the return of the G band doublet, E2g2(i) (1573 cm−1) and E2g2(b) (1601 cm−1), as well as the returning 2D(2) band, between ca. 0.17–0.22 V; and eventually, the reappearance of a sharp singlet G band at 1586 cm−1 and the 2D(1) band, and subsequent band shifting back to the original OCP band positions between ca. 0.3–1.5 V.
E (V) | G | E2g2(i) | E2g2(b) | 2D(1) | 2D(2) | |||||
---|---|---|---|---|---|---|---|---|---|---|
ω (cm−1) | FWHM (cm−1) | ω (cm−1) | FWHM (cm−1) | ω (cm−1) | FWHM (cm−1) | ω (cm−1) | FWHM (cm−1) | ω (cm−1) | FWHM (cm−1) | |
3.00 | 1580 | 12 | — | — | — | — | 2625 | 61 | 2688 | 53 |
0.76 | 1580 | 14 | — | — | — | — | 2625 | 57 | 2687 | 50 |
0.53 | 1581 | 9 | — | — | — | — | 2632 | 68 | 2687 | 36 |
0.35 | 1586 | 7 | — | — | — | — | 2622 | 30 | 2683 | 35 |
0.24 | 1589 | 9 | — | — | — | — | 2615 | 27 | 2681 | 45 |
0.2 | 1590 | 11 | — | — | — | — | — | — | 2656 | 75 |
0.19 | — | — | 1576 | 12 | 1599 | 15 | — | — | 2646 | 73 |
0.18 | — | — | 1575 | 9 | 1600 | 15 | — | — | 2629 | 59 |
0.15 | — | — | 1574 | 11 | 1601 | 15 | — | — | 2611 | 31 |
0.1 | — | — | — | — | — | — | — | — | — | — |
0.086 | — | — | — | — | — | — | — | — | — | — |
0.069 | — | — | — | — | — | — | — | — | — | — |
0.035 | — | — | — | — | — | — | — | — | — | — |
0.14 | — | — | — | — | 1592 | 57 | — | — | — | — |
0.15 | — | — | — | — | 1598 | 44 | — | — | — | — |
0.17 | — | — | 1573 | 8 | 1601 | 30 | — | — | 2614 | 51 |
0.22 | — | — | 1573 | 11 | 1598 | 22 | — | — | 2636 | 73 |
0.23 | — | — | 1577 | 16 | 1597 | 23 | — | — | 2649 | 90 |
0.31 | 1586 | 11 | — | — | — | — | 2630 | 48 | 2682 | 45 |
1.1 | 1579 | 15 | — | — | — | — | 2635 | 60 | 2685 | 39 |
At ca. 0.20 V, accompanying the onset of stage 4 GIC formation, the gradient of the 2D(2) peak red-shift increases significantly to a rate of 802 ± 87 cm−1 V−1 (Fig. 6, Table 1). A similar red-shift in the 2D position has previously been observed during the n-type doping of, and the intercalation of other metal ions into GICs.24,26 This shift may be attributed to electronic doping and the increased in-plane (biaxial) lattice strain accompanying Li+ insertion. The linear nature of this shift between 0.24–0.15 V, and the biaxial nature of the attributed lattice strain, implies that the graphene layers in the graphitic carbon are becoming increasingly distorted on lithium intercalation. This is in keeping with the Daumas–Hérold model of intercalation, where the graphene layers are flexible and deform around the intercalating lithium ions, and contrary to the Rüdorff model, which proposes a sequential filling up of alternating graphene interlayer spaces with no structural distortions induced within the individual graphene sheets, as seen in Fig. 7.
Fig. 7 A schematic of the (A) Rüdorff and (B) Daumas–Hérold models of ion intercalation into graphite. |
The apparent shape change of the 2D band towards a single Lorentzian peak suggests the graphene layers in the stage 4 GIC have become electronically similar to single layer graphene, most likely due to electronic decoupling resulting from the transfer of charge from the electron from the external circuit to the bounding layers.24 The 2D band is no longer observable below 0.10 V. This suggests that stage 2 and stage 1 GIC are not visible in the Raman spectra as all graphene layers of the GIC are charged, and thus no longer give a Raman signal.
The phase transition is related to the change from a random distribution of lithium ions to a more compact distribution in the staged compound. The shift of the E2g2(b) band to 1601 cm−1 is due to a further increase of the C–C bond force constants.
This shift can then be explained by charge transfer effects during intercalation of a donor species where electrons occupy the π* orbital and thus weaken the C–C bond strength. Such an effect has been seen for the intercalation of donor species into less ordered carbon materials.13,35,36 This broad band continues to blue-shift and broadens until it disappears into the noise at ca. 1540 cm−1. Raman bands are thought to be no longer visible because of the increase of electrical conductivity of these low stage GICs, which leads to a reduction in optical skin depth, and results in a low Raman scattering intensity.7
From 0.22 to 0.24 V there is a mixture of stage 4 and dilute stage 1, and the doublet G band returns again to the single E2g2 band at 1586 cm−1, which then blue-shifts back to 1579 cm−1, showing the passage from dilute stage 1 to fully de-intercalated graphite. The shift in G band position is shown clearly between 0.31–1.1 V vs. Li/Li+.
The Raman spectra before and after lithiation are shown in Fig. 8. Both show very similar features, and the relative intensities of the G and 2D bands remain the same. Both spectra display a weak D band, indicating that the area measured did not become further disordered during the first Li+ insertion/extraction cycle. The lack of change in the spectra indicates that the measured graphite particle has experienced no significant structural change.
Fig. 8 In situ Raman spectra before and after lithiation, * indicates signals from electrolyte bands. |
The reappearance of the 2D band is discernible at 0.17 V at 2614 cm−1 with a single Lorentzian fit that we have ascribed to the returning 2D(2) peak. This is a similar position to the last observable position during lithiation (2611 cm−1 at 0.15 V) suggesting the return of the stage 3 GIC. The 2D(2) peak red-shifts and broadens between ca. 0.17 V to 0.23 V until an eventual reforming of the low energy 2D(1) peak at 0.31 V. This splitting of the 2D band marks the renewed electronic coupling between the graphene layers and suggests the formation of the dilute stage 1 GIC. Both 2D(1) and 2D(2) peaks shift back towards their OCP positions.
The obtained first order spectra agree with previous reports, in particular with the observation of the G-band splitting, indicating the formation of a staged GIC. The behaviour of the second order in situ Raman spectra was reported for the first time during the first lithium insertion and extraction cycle. The 2D band was observed to shift to lower wavenumbers (2681 to 2611 cm−1) when stage 4 and 3 graphite intercalation compounds were formed. The shift of the 2D band supports the Daumas–Hérold model of electrochemically derived intercalation, where the graphene layers are flexible and deform around domains of ordered intercalated lithium ions. The opposite trends were observed in the Raman spectra during lithium extraction, indicating the lithium is removed in a reversible manner, with comparable initial and final spectra. This demonstrates that, at this particular point of the graphite electrode, no disordering of the structure resulted from a single cycle of lithium insertion and extraction.
This journal is © The Royal Society of Chemistry 2014 |