Simple Way of Making Free-Standing Battery Electrodes and their Use in Enabling Half-Cell Impedance Measurements via μ-Reference Electrode

The following describes an easy approach to produce free-standing PVDF-bonded electrodes. Other methods to produce free-standing electrodes can be found elsewhere.^17,18 The two here proposed methods (acc. to procedures A or B, see Fig. 1) are suitable to produce free-standing PVDF-bonded electrodes of both graphite and NCM (or other cathode active materials), whereby for the latter only procedure A can be used, as electrodes with cathode active materials generally should not be immersed in water. The first step in preparing free-standing electrodes is coating a thin layer of sodium carboxymethyl cellulose (CMC, Sigma Aldrich) binder onto a current collector foil. The CMC binder was mixed at a CMC to H₂O ratio of 1:25 (wt:wt) and coated at a wet film thickness of 30 μm onto a copper current collector foil (MTI, 12 μm) attached to a glass plate using a gap bar coater (RK PrintCoat Instruments, UK). To prepare a free-standing graphite electrode, graphite (T311 type from SGL, with 19 μm D50 and 3 m² g⁻¹) and polymer binder (polyvinylidene fluoride (PVDF), Arkema) at a ratio of 95:5 (wt:wt) were mixed with N-Methyl-2-pyrrolidone (NMP, Sigma Aldrich, anhydrous, 99.5%) at a solid to liquid ratio of 5:4 (wt:wt) in a planetary mixer (Thinky ARV-310) at 2000 rpm for five minutes. The prepared graphite slurries were coated onto the CMC-coated copper current collector at a wet film thickness of 200 μm. Free-standing cathode electrodes can be prepared analogously.

Zoom In Zoom Out Reset image size

Since the CMC layer is not soluble in NMP, it serves to block the PVDF binder of the electrode coated on top of it from adhering to the current collector foil. The dried electrode is then put into a bath of ethanol (see panel A1 in Fig. 1, left-hand-side) or water (see panel B1 in Fig. 1, right-hand-side), whereby the latter is only recommended for materials stable in H₂O. Since CMC is not soluble in ethanol, the CMC film is not dissolved, but its surface becomes non-adherent and the electrode can be peeled or swiped off the surface with little effort (see panels A1-3 in Fig. 1). For graphite electrodes, the electrode sheet can also be immersed in a water bath in which the CMC layer dissolves, leaving the electrode detached from the surface, which can also easily be done with already punched electrodes (see panels B1-3 in Fig. 1). As CMC is commonly used in battery electrodes, residual CMC in the electrode is of little concern in battery tests; however, if necessary, it could be removed in an additional washing step. Additionally heating the water bath is possible to speed up the dissolution process of the CMC but no heat treatment was performed for this work. For the free-standing graphite (FSG) electrodes used in this study, only procedure A was used.

The thickness of the here prepared dried and uncompressed free-standing graphite electrodes and of conventional graphite working electrodes (prepared as above, but without the CMC interlayer) was ∼105 μm (±2%), corresponding to graphite loadings of ∼9.8 mg cm⁻² (±2%) with an areal capacity of ∼3.4 mAh cm⁻² (calculated for a theoretical graphite capacity of 350 mAh g⁻¹). The electrodes were not further compressed/calendered, and their porosity was ∼55% (based on thickness and areal weight measurements). The dried electrodes were punched out to a diameter of 10.95 mm (equating to an area of ∼0.94 cm²) using an electrode punch (Hohsen Corp. OSAKA, Japan). Alternatively, a carbon paper (5% hydrophobized ∼360 μm thick Toray carbon fiber paper T120, Toray) was used as free-standing electrode and applied as received. The Carbon paper has a roughness factor of ∼18 cm²_BET cm⁻², based on ∼7 mg_carbon cm⁻² areal weight and ∼0.25 m² g⁻¹ BET area while the FSG electrode has a roughness factor of ∼300 cm²_BET cm⁻², based on ∼9.8 mg_graphite cm⁻² areal weight and ∼3 m² g⁻¹ BET area.

The equivalent circuit for the impedance measurements shown in Fig. 2b was soldered according to the circuit description of Fig. 2a, using carbon film resistors (Conrad Electronic GmbH, Germany) and electrolyte (≥1 μF) or film (1 nF) capacitors (Conrad Electronic GmbH, Germany).

Zoom In Zoom Out Reset image size

For electrochemical impedance analysis, a three-electrode cell setup (Swagelok^® T-cell) with a gold-wire reference electrode (GWRE, described in more detail in Fig. 1b in Ref. 8) was used. The cells were built inside an argon filled glove box (MBraun, 25 °C ± 1 °C, oxygen and water content <0.1 ppm, Ar 5.0, Westfalen). All cell parts were dried at 120 °C in a vacuum oven (Büchi, Switzerland) for 8 h before being transferred into an Ar-filled glovebox.

The cells were assembled with a graphite working electrode (areal capacity ∼3.4 mAh cm⁻²), two porous glass fiber separators with a diameter of 11 mm (VWR, 250 μm uncompressed thickness, 90% porosity), and a counter electrode consisting of either a metallic lithium foil (0.45 mm thickness and 11 mm diameter, Rockwood Lithium) or a free-standing graphite electrode firmly attached to the metallic lithium foil (referred to as Li/FSG). For cycling stability tests and impedance analysis of the Li-metal or the Li/FSG electrodes, symmetric cells in a three-electrode cell setup as described above were assembled. 80 μl of LP57 electrolyte (1 M LiPF₆ in EC:EMC 3:7 (wt:wt), battery grade, BASF) were added to the cells. Using a potentiostat (Bio-Logic Science Instruments, France), the gold-wire reference electrode was lithiated at 150 nA for 1 h via the Li-metal or Li/FSG counter electrode in a temperature-controlled chamber (25 °C, Binder).

The cycling protocol started with a 3 h open circuit voltage phase to allow for complete wetting of the electrodes and stabilization of the Li/FSG counter electrode. Lithiation of the graphite working electrode was performed galvanostatically (constant current lithiation) at C/10 to 40 mV vs Li⁺/Li without prior formation. Potentiostatic electrochemical impedance measurements over the course of cycling were performed at open circuit voltage (OCV) at 100% state-of-charge (SOC) from 30 kHz to 0.1 Hz and with an excitation of 10 mV. For symmetric cell measurements, graphite electrodes prepared as mentioned above were charged to 40 mV, harvested, rinsed three times with 0.1 ml of diethylene carbonate (BASF SE, Germany), and re-assembled into symmetric cells using the above described assembly protocol but without a GWRE.

There are multiple sources of error for impedance measurements, ranging from the potentiostat to artefacts that may arise from an individual cell design or cell chemistry.^11–15 To exclude any source of artefacts that may be caused by the three-electrode T-cell setup or from electrochemical phenomena in the half-cell, the cell setup was mimicked as a model equivalent circuit using actual resistors and capacitors, allowing us to probe whether the potentiostat controls circuitry might lead to impedance artefacts.

Figure 2a shows the simplified equivalent circuit that represents a lithium-ion battery half-cell with a gold wire μ-reference electrode. Each of the three electrodes is represented by a parallel resistor/capacitor element (further on referred to as R/C element) with approximate resistance and capacitance values that represent those of a lithium-ion battery half-cell composed of a μ-RE, a Li-metal CE, a conventional graphite WE, and a separator:

• (1)  
  The resistance and capacitance of the here used μ-RE (the GWRE) were measured to be on the order of 10⁶ Ω and 1 nF, respectively, values which were used in the REF-branch of the model equivalent circuit (see top branch in Fig. 2a) and which can be rationalized by the very small GWRE cross sectional area of ∼20 × 10⁻⁶ cm² (this equates to a reasonable GWRE charge transfer resistance of 20 Ω·cm²).
• (2)  
  The Li-metal CE resistance was measured to be on the order of 10² Ω (based on Li/Li symmetric cell measurements, see Fig. 3a, red and green lines) and its capacitance was measured to be on the order of 1 μF (calculated from the semicircle apex at 1077 Hz of the Li-metal impedance response, red line Fig. 3a); both values are consistent with previous measurements.⁸ Thus, the capacitance for the Li-metal CE part of the model equivalent circuit (left branch in Fig. 2a) was set to 1 μF, while its resistance was set to R_CE = 220 Ω. However, in order to investigate the effect of a lower CE resistance on the impedance response of the model equivalent circuit, we also conducted experiments with a resistor of R_CE = 10 Ω.
• (3)  
  The graphite working electrode resistance was approximated to be on the order of 5 Ω (see right branch in Fig. 2a), based on previous measurements with the same graphite material which showed overall graphite anode low-frequency resistances of ∼2-4 Ω^8,10 for ∼1.5-fold lower loadings. The capacitance was set to 1 mF, representing the large surface area of a standard battery electrode (the frequency of the data point closest to the apex in Fig. 2b [dashed black and dotted blue line] is 35 Hz).
• (4)  
  Finally, the commonly observed overall resistance for the separator of ∼2 Ω¹⁰ was divided into two 1 Ω resistances (see center part of the circuit in Fig. 2a).

The model equivalent circuit (Fig. 2a) was measured with two different resistance values for the counter electrode, namely with R_CE = 220 Ω to represent the typical value of a Li-metal CE (see above), and with a substantially lower value of R_CE = 10 Ω. Figure 2b shows the resulting Nyquist plots for the WE impedance measured via the reference electrode terminal (i.e., the RE cable of the potentiostat is connected to the contact marked REF in Fig. 2a; WE and CE cables are connected to the contacts marked WE and CE, respectively). For R_CE = 220 Ω (red line), the WE impedance response (red line in Fig. 2b) shows an "overshoot" for frequencies below 110 Hz if compared to the expected semi-circle shape obtained when directly measuring the R/C circuit representing the working electrode (dashed black line; here, the RE and CE cables of the potentiostat are connected both to the contact marked REF 2 in Fig. 2a). Furthermore, a non-physical "back-looping" of the low-frequency branch is observed towards the lowest measured frequencies (i.e., near 0.1 Hz; see red line). Surprisingly, lowering the CE resistance to R_CE = 10 Ω gives the correct response of the impedance measurement (compare dotted blue and dashed black line). This is unexpected, as the magnitude of the counter electrode impedance should not influence the impedance response of the WE measured via a reference electrode. The exact reason for this artefact is not known, but we presume this to be an effect caused by the large impedance of the reference electrode in combination with the unknown internal impedances of the potentiostat, similarly to the descriptions in Ref. 12 Other explanations for the existence of impedance artefacts relating to the cell geometry and the reference electrode location^12–15 can be excluded, as the measurement on the model equivalent circuit is not affected by such influences. Hence, the correct impedance response with a high impedance μ-RE can be obtained by decreasing the value of the CE resistance, so that it is of a similar order of magnitude as that of the WE.

The following section describes an approach to minimize and stabilize the impedance response of the Li-metal CE in order to allow for artefact-free EIS measurements of the WE using a μ-RE in a half-cell setup.

The impedance artefacts observed with the model equivalent circuit explain why artefact-free impedances can be obtained in full-cells with a μ-RE,^8–10 as there the WE and the CE have comparable impedances. On the other hand, it also explains why we so far had not been able to acquire artefact-free WE impedances in half-cells with a Li-metal CE, as there the impedance of the WE is typically 10¹−10² times larger than that of the CE (see estimates above). The latter is illustrated by the graphite WE impedance response obtained for a half-cell with a GWRE and with a Li-metal CE (red line in Fig. 2d; cell configuration shown in the left panel of Fig. 2c), where a back-looping of the Nyquist plot near the low-frequency region occurs, rather similar to what was observed for the model equivalent circuit when R_CE ≫ R_WE (red line in Fig. 2b).

As mentioned in the introduction, half-cell measurements with a Li-metal CE provide valuable insights when evaluating new/modified anode or cathode active materials as working electrode, since the effect of a loss of cyclable lithium can be avoided by the large lithium excess and since the CE potential is well-defined (at least at low current densities, i.e., at low charge/discharge rates). However, in order to also enable an evaluation of the WE impedance with a μ-RE (i.e., in-situ, without the need to reassemble symmetric cells), the impedance of the Li-metal CE would need to be substantially reduced. This can be achieved by attaching a free-standing graphite electrode to the Li-metal surface, according to the cell configuration shown in the right-hand panel of Fig. 2c. Once the cell is filled with electrolyte, the free-standing graphite in contact with the metallic lithium (Li/FSG) is immediately lithiated and reaches a stable potential of 0 V vs Li⁺/Li within ∼3 h. At the same time, the largely increased interfacial surface area of the Li/FSG vs. the Li-metal CE leads to a drastic reduction of the CE impedance (as will be shown below), which should enable the acquisition of artefact-free WE impedances in a half-cell with a μ-RE and an Li/FSG CE. That this is indeed the case is shown by the graphite WE impedance response when using a GWRE and a low-impedance Li/FSG CE, and by comparing it to the symmetric cell measurement (blue and black line in Fig. 2d), where the back-looping of the Nyquist plot near the low-frequency region does not anymore appear, as expected on the basis of the model equivalent circuit experiments (blue line in Fig. 2b). This demonstrates that artefact-free in-situ WE impedance data can be obtained in a half-cell configuration, when a FSG is attached to a conventional Li-metal CE to lower its impedance. The apex frequency values of the measured graphite impedances for all three measurements in Fig. 2d (around 1 kHz) are more than one order of magnitude higher than the apex of the equivalent circuit measurements in Fig. 2b (around 35 Hz). This is the result of the more complex impedance of the graphite electrode which does not only include the charge transfer resistance, but also the pore resistances from the electrolyte within the pores of the graphite electrode as well as the solid electrolyte interphase. A detailed analysis of the graphite impedance will be shown in a later publication. In a recent work, we had also shown that the free-standing electrode concept can equally well be applied to determine the working electrode impedance in sodium-ion battery half-cells, using carbon paper attached to metallic sodium as the counter electrode.¹⁹

As the resistance of the CE is an important factor in enabling artefact-free EIS measurements of the WE via a μ-RE, we compared three different CE designs in order to give an overview over the range of resistance for different electrode designs: (i) lithium metal (referred to as Li-metal), (ii) carbon paper attached to Li-metal (referred to as Li/C-paper), and (iii) a FSG electrode attached to Li-metal (referred to as Li/FSG).

In order to obtain estimates for the impedance of a Li-metal electrode, a Li/FSG electrode, and a Li/C-paper, symmetric cells (i.e., Li//Li, Li/FSG//Li/FSG, and Li/C-paper//Li/C-paper cells) equipped with a GWRE were assembled. The impedance response (potentiostatic at OCV) of the working electrode (i.e., one of the two electrodes in the cell) was recorded using the GWRE. As shown in Fig. 3a, the low-frequency resistance (LFR, corrected by the high-frequency resistance (HFR)) of the pristine Li-metal electrode is ∼200 Ω (red lines) and gets reduced substantially to ∼40 Ω (green lines) after 5 h of Li-stripping at 0.32 mA cm⁻² (for the graphite WE used in this study, this would correspond to a C-rate of ∼0.1 h⁻¹). We attribute the drastic change in the lithium metal impedance to the change in lithium surface area after stripping, highlighting why cell impedance measurements of half-cells with a Li-metal CE and without a reference electrode do not allow to quantify the impedance of the WE. Figure 3b shows the impedance response of a Li/C-paper electrode (black line), which is a quick and easy alternative to a free-standing graphite electrode, as it is commercially available and can be applied as-received. While its LFR is lower than that of Li-metal (∼25 Ω), it is still ∼7 times larger compared to that of the pristine Li/FSG electrode (∼3–4 Ω; see magenta lines in Fig. 3c), in part due to the roughly one order of magnitude smaller surface area of the C-paper (see experimental section).

Zoom In Zoom Out Reset image size

Owing to the very low impedance of the Li/FSG electrode, the presence of a contact resistance (R_contact) can be discerned at high frequencies, which must be due to an imperfect lamination between the FSG and the Li-metal, so that the LFR for the Li/FSG electrode contains contributions from this contact resistance. Nevertheless, the overall LFR of the Li/FSG electrode is roughly an order of magnitude lower than that of Li-metal and rather similar to the LFR of a conventional graphite electrode (∼8 Ω, see black line in Fig. 2d).

In order to evaluate the variation of the Li/FSG electrode impedance with cycling, the Li/FSG//Li/FSG cell was additionally cycled 10 times at ±0.21 mA cm⁻² for 10 h (10 h stripping and 10 h plating per cycle). The blue line in Fig. 3c shows the impedance response after 10 cycles. The electrode impedance is slightly larger, with an additional shift to higher values attributed to an increase in electrical contact resistance between the FSG and the Li-metal (this could likely be suppressed by a higher cell compression, which in the current experiments is ∼2 bar, set by a spring).⁸ After 5 additional cycles at ±0.42 mA cm⁻² for 5 h (i.e., 5 h stripping and 5 h plating per cycle), the impedance again shifts to the right, attributed to an additional increase in contact resistance. Overall, the Li/FSG electrode shows a slow increase in impedance for low current cycling; its impedance is independent of the state-of-charge of the working electrode, since on account of the quasi unlimited Li-supply from the Li-metal, the Li/FSG potential remains at 0 V vs Li^*/Li at OCV, where the impedance data are acquired. In this configuration, the WE impedance can in principle be estimated even from cell impedances acquired for half-cells with an Li/FSG CE and without a μ-RE over the course of a few cycles, if the impedance response of the Li/FSG electrode was measured independently beforehand, e.g., in symmetric cells. In this case, any reversibly measured changes in cell impedance can be attributed to changes of the working electrode impedance, allowing the user to extract useful information from a half-cell impedance measurement even without reference electrode. Such a setup was not used in this publication and it is of the discretion of the individual reader to determine its validity for individual setups.

As shown in this publication, the choice of CE affects the ability to perform artefact-free impedance analysis of the WE via a μ-RE. Two options of free-standing electrode to minimize the CE impedance and to enable the measurement of the WE impedance were presented here. First, the free-standing graphite electrode (FSG) attached to a Li-metal electrode (Li/FSG) shows the lowest impedance and is therefore best suited to allow the use of a μ-RE for any commercial Li-ion battery active materials known to the authors. This is true for a free-standing electrode of a reasonable loading of 1–3 mAh cm⁻² comprised of commercially available graphite for Li-ion batteries. Its drawback, however, are the additional steps needed to produce the electrode and during cell assembly.

Using a carbon paper in contact with Li-metal (Li/C-paper) is more convenient, as carbon papers are commercially available and can be used as received. The drawback of the Li/C-paper configuration is its generally higher impedance (depending on the type of carbon paper), which is why it may not be suited for measurements of all types of working electrodes.

In all cases, however, the individual user needs to know the resistance range of the working electrode to be examined and should, for best practice, ensure that the counter electrode resistance is always equal or lower in resistance compared to the WE as the impedance ratio of WE and CE that marks the onset of artefacts is not know and may depend heavily on the cell setup and potentiostat.

Regarding the choice of Li-metal, any commercially available Li-metal films with thickness of 75 μm and above are sufficiently overbalanced to serve as Li-source for this setup.