Probing Structural Evolution and Charge Storage Mechanism of NiO2Hx Electrode Materials using In Operando Resonance Raman Spectroscopy

In operando resonance Raman spectroscopy suggests quantitative correlation between phonon band properties and the amount of charge storage of high‐energy density NiO2Hx battery/pseudocapacitive material. Comparing the spectroscopic evolution using different electrolytes reveals the contributions of breaking/formation of O–H bonds and insertion/extraction of cations to electrochemical charge storage of NiO2Hx.


Raman spectroscopic measurement
Raman spectra were obtained using a Renishaw RM 1000 spectromicroscopy system with a 20x/0.40 objective. An air-cooled Ar ion Laser (Mellos Griot) with wavelengths of 514 nm and 488 nm and a He-Ne Laser (Thorlab HRP-170) with wavelength of 633 nm were used as excitation of Raman spectra. The confocal slit was adjusted to be 5 um to minimize the band broadening effect due to the contribution of non-confocal signal.

Reflection spectroscopic measurement
The reflection spectroscopic measurement was performed using a reflection integration sphere ). An Ocean Optics DH-2000 light source, QP450-1 optical fiber, and a DH-2000 spectrometer were applied as light source, light transmission fiber, and detector respectively. The reference spectrum was acquired using an Ocean Optics reference standard.
The acquisition time was 65 s with boxcar value of 3 to smooth the spectrum. The absorption spectrum of the NiO 2 H x model electrode was obtained by comparing the reflection spectra of both reference standard and model electrode.
Also, the reflection spectra measurement was conducted through customization of Raman spectrometer for better signal-to-noise ratio. A halogen lamp was used for microscope white light illumination. The reflected light was collected by a high NA objective (50X/0.75) and was then guided to grating chamber. The absorption spectrum of the NiO 2 H x model electrode was also calculated by comparing the reflection spectra of both Ocean Optics reflection standard and the model electrode.

In operando Raman cell setup
An in-operando Raman cell was used to collect Raman spectra and perform electrochemical measurement at the same time. The detailed design of the in operando cell is shown in Figure S1 and was described in our previous work. [S1] The cell body was made of Teflon PTFE with two series of concentric holes with different diameters. A stainless steel rod wrapped by PTFE insulation tape (to prevent leakage) which was inserted to the bottom of the cell served as the electric connection of WE. A Pt mesh connected with a Pt wire was applied as counter electrode.
The Whatman filteres (GF/D) were used as separators to separate WE and CE. Similar with CE, a reference electrode (FlexRef WPI instruments) which was wrapped by PTFE tube was inserted into the other side of the cell. The tip of the RE was located exactly near the edge of WE and CE.
Both of the RE and CE were wrapped by PTFE tape to prevent leakage. On the basis of threeelectrode configuration, a quartz window was adhered on a PTFE washer on the top of the cell.
The cell cap was bolted on the cell body with PTFE O-rings to prevent evaporation of electrolyte. Figure S1. Construction of the three-electrode in operando Raman cell.

Electrochemical measurements
During in operando Raman spectroscopic acquisition, cyclic voltammetry (CV) measurements were performed at room temperature using an electrochemical workstation (Solartron, SI 1285) with a scan rate of 10 mV s -1 for 12 cycles. Raman spectroscopic acquisition time for each spectrum is 5 s (including the time for CCD camera response) and each Raman spectrum is acquired within 0.05 V consequently. The Raman spectrum of each potential in different cycles (totally 12 cycles) were then accumulated to improve the signal to noise ratio.
The charge-discharge experiments with different current densities were performed also using Solartron SI 1285. CV measurements under different scan rates was conducted using an electrochemical workstation (PARSTAT MC 1000).

Formation of NiO 2 H x thin film model electrode
When the bare Ni foil was immersed in 2 M KOH electrolyte, a thin layer of nickel hydroxide will be formed on the surface, because Ni metal starts to be unstable as long as the potential of Ni is higher than -0.7 V vs Ag/AgCl in strong base solution according to the Pourbix diagram of Ni/NiO 2 H x /NiO 2 . [S2, 3] During extensive cycling of the Ni foil between 0 V and 0.6 V, the foil surface will be gradually oxidized to form γ-NiOOH eventually. [S3-5] Figure S2 shows the CV profile of the electrochemical oxidation of Ni foil within 3000 cycles. The foil exhibited negligible current density in the first cycle. However, it is obvious that the current density gradually increases with cycles numbers and doesn't change significantly when the cycle number approach 3000. It is also noted that the over potential of the redox peak gradually increase with current density, which is due to the increased thickness of the nickel hydroxide film caused higher ohmic resistance. Figure S2. CV profiles of Ni foil within 3000 cycles with an interval of 50 cycles. The scan rate was 50 mV s -1 .

SAED pattern of NiO 2 H x thin film model electrode
Due to the limited thickness of the NiO 2 H x film, the X-ray diffraction pattern of the film barely exhibited diffraction signal of NiO 2 H x . Therefore, we applied TEM for the phase analyses. Figure 1c shows the TEM image of the scraped thin film of NiO 2 H x , which exhibits clear layered fringes. The SAED pattern is shown in Figure S3 which indicates three weak diffraction rings.

Raman bands assignment
As mentioned in the main text, the layer-structured NiO 2 H x has complicated structure and stoichiometry, regarding to the amount of the bonded hydrogen and interlayer ions. Depends on the bond length and coordination of oxygen, the space group of the NiO 2 H x complex was reported as a variety of types (e.g. hexagonal R -3m or monoclinic C 2/m ). [S4, 6-8] In order to simplify the phonon band assignment, the simple model of NiO 2 layer with highest order of symmetry (R -3/m ) was applied, which doesn't take the bonded hydrogen and interlayer ions into account ( Figure S4). Figure S4. Illustration of simplification of NiO 2 layer structure model for Raman band assignment based on group theory.
Based on group theory analyses, the irreproducible representation of the NiO 2 layer can be written as: (1) One A 1u and one E u mode are acoustic modes. Among the four optical modes, the A 1g and E g mode are Raman active, which basically match the experimental results that only two Raman bands were observed. For the A 1g mode, the oxygens vibrate perpendicular to the plane formed by oxgyen; whereas the oxygens vibrate along this plane for the E g mode (Figure 1f). The Raman tensors of the two modes are: (2) It is noted that the Raman tensor of A 1g mode is diagonal matrix, whereas the Raman tensor of E g mode has non-diagonal elements. Therefore, the scattered light of the A 1g mode will largely maintain the polarization direction of the incident laser and the scattered light of the E g mode will be depolarized due to the polycrystalline nature of the film. By switching the polarization configuration from Z(XX)Z to Z(XY)Z , the relative intensity of the A 1g mode is greatly reduced, confirming the Raman band assignments and the validity to apply the simple NiO 2 layered model to approximate the structure of NiO 2 H x .

Absorption spectrum of NiO 2 H x model electrode
The absorption spectrum of NiO 2 H x model electrode was obtained by comparing the reflection spectra of a reference standard and the model electrode. Since there is no light transmission for the NiO 2 H x model electrode, the absorbance can be expressed by the following equation (R is reflectance):

⁄
(3) Figure S5 shows the reflection spectra of both reference standard and the model electrode   Since the reflection spectra acquired by the customization of Raman spectrometer was collected by an objective (50x/0.75), a small portion of reflected light cannot be collected. Therefore, a standard Ocean Optics UV-vis spectrometer equipped with an integration sphere was applied to the measurement of the reflection spectra. Due to the limited efficiency of integration sphere and much lower sensitivity of CCD detector of UV-vis spectrometer than that of Raman spectrometer, the signal-to-noise ratio of the absorption spectra ( Figure S6b) acquired by Ocean Optics is much lower than that of the spectra acquired by customization of Raman spectrometer ( Figure S6a). However, it is obvious that the two spectra show same absorption characteristics generally, which is the broad absorbance profile with a maximum near 700 nm.

Raman spectra band fitting
The Raman fitting calculation of each Raman spectra were done by assuming a linear baseline and that all Raman bands have a Lorentzian line shape: In this equation, is the Raman shift; I is Raman band intensity, and Γ are position and FWHM of Raman band, respectively. The band fitting process was conducted through a Matlab fitting code automatically. Since only two bands were fitted (A 1g and E g ) and the positions of the two bands are well-separated, the automatic band fitting can ensure the accuracy for band evolution analysis.

The measurement of the reflectance under in operando conditions
In order to corroborate the analyses obtained from in operando Raman spectroscopic measurment, we performed the measurement of reflectance of the model electrode under in operando conditions. Light reflection is complementary to light absorption, which corresponds to the electronic structure of the materials (i.e. reflecting the oxidation state). Figure S7 shows the reflectance profile of the electrode as the potential (vs Ag/AgCl) of the working electrode (WE) was changed from 0 V to 0.5 V in a 2 M KOH electrolyte solution. A signficant red shift of the reflection profile (correpsonding to a blue shift of absorption) is clearly observed as WE potential was switched from 0 V to 0.5 V. This shift suggests a increase in the energy gap of electronic state transition and evolution of the oxidation state of Ni, consistent with the Raman spectroscopic evolution as a function of electrochemical potential. It is noted that the reflection profile under the in operando condition shown in Figure S7 exhibited discrepancies with the light reflectance profile under ex-situ condition shown in Figure S5, since the existence of electrolyte will strongly interfere with light absorption/reflection. Also, the light reflectance profile under in operando condition cannot be converted to the light absorption profile due to lack of the light reflectance behavior of a reference standard under the same operando condition.
However, the evolution of light reflectance alone can provide strong evidence of change in oxidation state during cell operation.  Figure S8 shows the CV profiles of NiO 2 H x model electrode tested in 2 M KOH electrolyte with different scan rates (the potential window for 500 mV s -1 was expanded). It is obvious that the CV profile was well-maintained under high rate operation. Similarly, the charge-discharge experiment with different current densities was also performed to calculate the capacitance/capacity retention ( Figure S9 and S10). At a high current density of 5 mA cm -2 (427 A g -1 with respect to the mass loading of the active material; a charge/discharge cycle is finished within 5 seconds), the capacitance/capacity can be maintained more than 80 %. These experimental facts indicate the flat geometry of NiO 2 H x model electrode can greatly reduce the kinetic hindrance caused by electrolyte mass transport, which is crucial for unambiguous and quantitative correlation between phonon properties and electrochemical features. As suggested by Simon et al. and Brousse et al., [S9, 10] the electrochemical chemical behavior with distinctive and abrupt CV peaks belongs to a typical battery behavior. Therefore, the units for the charge storage of battery system (Ah or C) were used for charge storage capacity calculation. The calculated areal and specific capacity are 10.98 mC cm -2 and 261 mAh g -1 (or 1,877 F g -1 ), respectively (considering the fact that a large number of reported works about battery-like materials have already used Farad as unit, the charge storage capacity calculated using the unit "F g -1 " is 1877 F g -1 ). Figure S9. The charge-discharge profiles of NiO 2 H x model electrode with different current densities in a 2 M KOH aqueous electrolyte. The CV profile NiO 2 H x model electrode tested in 2 M KNO 3 shows a rectangular-like shape and can be well maintained under high rate operation (500 mV s -1 , Figure S11), which is similar as most of cation incorporation-based pseudocapacitive charge storage (e.g. MnO 2 ). [S1] Figure S11. Cyclic voltammogram (CV) of NiO 2 H x thin film model electrode with different scan rates in a 2 M KNO 3 aqueous electrolyte.

Calculation of charge density as a function of electrochemical potential
The charge densities as a function of electrode potential shown in Figure 4 were integrated from the CV curves shown in Figures 2 and 3. The relationship between the CV current and the stored charge at a particular time (t 0 ) is expressed by the following equation.
where Q is the stored charge density in C cm -2 , i(t) is the CV current density in A cm -2 .
The CV profile shown in Figure 2 and the integrated charge density (shown in Figure 4d) are compared in Figure S12. At the highest electrochemical potential, the stored charge is maximized. Figure S12. The comparison of the CV profile shown in Figure 2 and the charge density shown in Figure 4d.

In operando Raman spectroscopic evolution under galvanostatic charge/discharge
In order to compare the spectroscopic evolution of NiO 2 H x under the conditions for cyclic voltammery shown in Figure 2, similar in operando experiments were conducted under galvanostatic charge/discharge. In cyclic voltammetry, the rate of electrochemical reations increases with the potential and reaches the maximum at the electrochemical potential where the redox peaks are observed. In contrast, the rate of electrochemical reactions is relatively constant under galvanostatic charge/discharge; the structural changes induced by electrochemical reacitons are more gradual. Figure S13 shows the in operando Raman spectroscopic evolution acquired at a constant current density of 50 µA cm -2 . As we expected, the evolution of Raman spectra was more gradual under galvanostatic charge/discharge conditions, not abrupt as seen under CV conditions, implying that the electrochemical current leads to the structural changes. Figure S13. The Raman spectroscopic evolution of NiO 2 H x model electrode acquired at a constant current density of 50 µA cm -2 in a 2 M KOH electrolyte between 0 V and 0.5 V.

Effects of electrolyte cations
In order to evaluate the effects of the types of the cations on the electrochemical behavior and the structural changes during cycling of the NiO 2 H x model electrode, we performed the same electrochemical and in operando Raman measurements on the NiO 2 H x electrodes from the same batch as the electrolyte solution was changed from KOH to NaOH and LiOH. Since the interlayer spacing of the NiO 2 H x i te i nges t 7 Å, [S11] intercalation of different types of cations such as Li + (0.59 Å), Na + (0.99 Å), and K + (1.37 Å) may influence the electrochemical behavior and charge storage mechanisms of the electrode. Figure S14 shows the CV profiles of the NiO 2 H x model electrode tested in 2 M aqueous electrolyte solution of LiOH, NaOH, and KOH. It is found that the evolution of in operando Raman spectra remained relatively the same when electrolyte of different type of cations was used, indicating that the charge storage mechanism (breaking/formation of O-H bond) is independent of the types of cations. However, the electrochemical behavior (e.g., CVs) of the NiO 2 H x model electrodes changed slightly as the type of electrolyte cations was changed from one type to another. For example, the redox peak positions in the CVs were shifted slightly to higher potentials as the electrolyte cations were changed from K + to Na + and to Li + , due most likely to stronger solvation effect for smaller cations. Figure S14, Cyclic voltammograms (CV) of NiO 2 H x thin-film model electrodes tested in 2 M aqueous electrolyte solution of LiOH, NaOH, and KOH. The scan rate is kept at 10 mV s -1 . Figure S15. Evolution of in operando Raman spectra of a thin-film model NiO 2 H x electrode cycled in 2 M aqueous KOH, NaOH, and LiOH aqueous electrolytes. The experimental setting is exactly the same as described in Figure 2. Figure S16 shows the comparison of the Raman spectra of as prepared NiO 2 H x model electrode and the NiO 2 H x model electrode immersed in 2 M KNO 3 electrolyte. It is obvious that the band profile changed considerably, especially the band width of ν 2 band (E g ). The band broadening is a clear indication of cation incorporation between the interlayer spacing, because the incorporated cations will lead to break of the lattice symmetry, leading to structural disorder and thus band broadening effect. [S1, 12-14] Similar band broadening effects related to cation incorporation has been reported in different works related to lithium ion battery and pseudocapacitive charge storage. [S1, 12-14] Therefore, Raman spectroscopic analyses clearly indicate that NiO 2 H x can incorporate cations, but the cation incorporation can't form massive charge storage proven by electrochemical and Raman spectroscopic analyses. Figure S16. Comparison of the Raman spectra of as-prepared NiO 2 H x model electrode and the electrode immersed in 2 M KNO 3 electrolyte at OCV condition.