Strategies to Improve the Performance of Carbon/Carbon Capacitors in Salt Aqueous Electrolytes

The commercial activated carbons DLC SUPRA 30 (Norit, S_BET = 2066 m² g⁻¹, V_micro = 1.04 cm³ g⁻¹, average micropore size L₀ = 1.16 nm) and YP 80F (Kuraray Chemicals Co, S_BET = 2270 m² g⁻¹, V_micro = 0.99 cm³ g⁻¹, average micropore size L₀ = 1.17 nm) were used for manufacturing pellet and coated electrodes, respectively. 9,10-phenanthrenequinone (Sigma-Aldrich, 95%) and tert-butyl nitrite (Sigma-Aldrich, 90%) were used for the modification of activated carbons. The aqueous electrolytes were based on lithium sulfate (Li₂SO₄.H₂O, Sigma-Aldrich, >99%), sodium sulfate (Na₂SO₄, Sigma-Aldrich, >99%), potassium sulfate (K₂SO₄, Avantor, 99%), magnesium sulfate (MgSO₄.7H₂O, Avantor, 98%), aluminum sulfate (Al₂(SO₄)₃.18H₂O, Avantor, 98%) and beryllium sulfate (BeSO₄.4H₂O, Sigma-Aldrich, >99%). Di-sodium hydrogen phosphate (Na₂HPO₄, Sigma-Aldrich, >99%) and sodium di-hydrogen phosphate (NaH₂PO₄, Sigma-Aldrich, >99%) were used to prepare buffer solutions.

Carbon modification was carried out by grafting according to the procedure described in reference³⁵ using spontaneous reduction of in-situ generated diazonium cations. 2-amino-9,10-phenanthrenequinone (186.3 mg, 0.1 equiv. compared to carbon) was added to acetonitrile (300 mL), and the solution was stirred until complete dissolution of the quinone derivative. Then, tert-butylnitrite (99 μL, 1 equiv. compared to quinone) was added to the solution, the mixture was stirred for 15 min and the DLC Supra 30 activated carbon (100 mg) was added. Thereafter, 1 equiv. of tert-butylnitrite was added two more times (after 30 minutes and 1 hour) directly to the organic suspension, and the mixture was kept at ambient temperature overnight under stirring. The modified carbon powder was finally collected by vacuum filtering on a nylon filtration membrane having a pore size diameter of 0.47μm (Pall) and washed by successive aliquots of acetonitrile, DMF, acetone and methanol in order to remove any unreacted species from the surface. Afterwards, the modified carbon was dried under vacuum overnight at 60°C; in the following part of the manuscript, this carbon will be referred to as DLC Supra 30-PQ.

The surface oxygenated functionality of as-received and grafted carbons was characterized by temperature-programmed desorption (TPD), using a TG equipment (TG209 F1 Iris, NETZSCH) coupled with a mass spectrometer (QMS 403C Aëolos, NETZSCH). In these experiments, ca. 6 mg of carbon powder was heated up to 950°C (heating rate 20°C min⁻¹) under helium flow rate of 50 mL min⁻¹. CO₂- and CO-evolving groups were quantified after calcium oxalate monohydrate calibration, taking into account CO disproportionation.³⁶

The porous texture of carbons was determined from nitrogen adsorption/desorption isotherms recorded at −196°C using an ASAP 2020 (Micrometrics). Prior to the measurements, the samples (around 60 mg) were degassed under vacuum for 12 h at 350°C for as-received carbons and 24 h at 110°C for grafted carbons. The pore size distribution (PSD) was determined using the quenched solid density functional theory (QSDFT),³⁷ and the total pore volume V_total as well as micro V_micro and mesopore volumes V_meso were obtained directly from the calculated cumulative PSDs. The average micropore size (L₀) was determined from the integration of the PSD area for pores below 2 nm. The static contact angle for as-received and grafted carbons was determined using a video-based optical contact angle measuring instrument OCA 15Pro (DataPhysics Instruments, Germany).

Self-standing pellet electrodes were prepared by mixing as-received or grafted (DLC Supra 30-PQ) carbon (80 wt%) with 10 wt% poly(tetrafluoroethylene) (PTFE - 60 wt% dispersion in water, Sigma-Aldrich) binder and 10 wt% carbon black (Super C65, Timcal) conductivity enhancer. The three components were mixed in isopropanol (Avantor, 99.5%) until a dough was obtained; then the dough was rolled into sheet and dried under vacuum at 110°C (as-received carbon) and 60°C (DLC Supra 30-PQ carbon) for 12 hours. 10 mm diameter (8–10 mg) pellet electrodes of thickness ∼0.2 mm were punched out from the sheet, and AC/AC capacitors were realized by sandwiching a porous glass microfiber membrane GF/A (Whatman, thickness = 0.26 mm) between two pellet electrodes and stainless steel 316L current collectors, using PTFE Swagelok-type vessels with or without inlet for a reference electrode. In order to avoid the issues related with conductivity as well as wetting with the electrolyte, electrodes of same mass and thickness were used. Hg/Hg₂SO₄ in 0.5 mol L⁻¹ H₂SO₄ was used as reference electrode for studies on as-received carbons, and a silver wire as quasi-reference electrode for cells with DLC Supra 30-PQ carbon. The solutions used in cells with as-received carbons were 0.5 mol L⁻¹ K₂SO₄ and 1 mol L⁻¹ Li₂SO₄, Na₂SO₄, MgSO₄, BeSO₄, and Al₂(SO₄)₃. The pH and conductivity values of these electrolytes are given in Table I. In case of cells built from the DLC Supra 30-PQ carbon, the electrolyte was a buffer solution, either 0.1 mol L⁻¹ phosphate buffer (pH = 7.2) in deionized water, or a commercial biphthalate buffer (pH = 4.0) purchased from Metrohm.

For realizing coated electrodes, the surface of grade 1.431 stainless steel (Interbelts, thickness = 15 μm) and nickel (Schlenk, thickness = 20 μm) foils was pre-coated with a thin layer (15 μm) of carbon conductive ink (Electrodag PF-407A, Acheson) to provide a rougher substrate. Activated carbon YP 80F (83.5 wt%), carbon black (Super C65, Timcal, 8.5 wt%) conductivity enhancer and polyvinylidene difluoride (PVdF, Kynar HSV 900, Arkema, 8 wt%) binder dissolved in 1-methyl-2-pyrrolidone (NMP, Sigma-Aldrich) were mixed with an homogeneizer (IKA Ultra-Turrax T 18 basic), and the obtained slurry was cast with a Doctor Blade applicator (Elcometer 3600) on the previously prepared surface of stainless steel or nickel foil. Afterwards, the coating was dried by slow evaporation in air overnight, followed by heating under vacuum at 120°C for 12 hours. 10 mm diameter disk electrodes (mass of active material ∼3.5 mg, coating thickness ∼100 μm) were punched out from the coated foil. AC/AC capacitors were realized in PTFE Swagelok-type vessel (with additional inlet for Hg/Hg₂SO₄ reference electrode) by sandwiching an absorptive glass mat separator (AGM, Bernard Dumas, thickness = 0.52 mm) between two coated electrodes and cylindrical current collectors, either from stainless steel or nickel. Before being closed, the assembled system was soaked under vacuum with 1 mol L⁻¹ Li₂SO₄ electrolytic solution.

The electrochemical performance of the two-electrode (without and with reference) and three-electrode cells was determined with a VMP3 (Biologic, France) multichannel potentiostat/galvanostat. The gravimetric capacitance C (F g⁻¹ of as-received as well as DLC Supra 30-PQ activated carbons in one electrode) was estimated from galvanostatic discharge at 0.2 A g⁻¹ and cyclic voltammetry at scan rates of 2 mV s⁻¹ and 5 mV s⁻¹. Accelerated ageing of the cells was realized as reported in Ref. 9 by a succession of two-hour potentiostatic floating periods, each followed by five galvanostatic (1 A g⁻¹ referred to the active mass of one electrode) cycles the cell potential hold during floating and the maximum cell potential during galvanostatic cycling were fixed at the same value. The capacitance and cell resistance were estimated from the fifth discharge to monitor the state of health (SOH) after each floating period. The test sequences were repeated 60 times, for a total floating time of 120 h. A digital pressure sensor Keller 35X Ei was connected to the capacitors in 1 mol L⁻¹ Li₂SO₄ or BeSO₄ to monitor gas evolution.

In this section, self-standing electrodes made from as-received DLC Supra 30 (further named AC) and stainless steel (grade 316L) current collectors were used. In our previous reports, we demonstrated that oxidation of the positive AC electrode and corrosion of the positive current collector are essentially the cause of performance deterioration during floating of AC/AC capacitors in 1 mol L⁻¹ Li₂SO₄ at 1.6 V.¹¹ According to the Nernst equation, the thermodynamic limits for di-oxygen and di-hydrogen evolution at the positive and negative electrodes, and consequently the possible performance degradation, are pH dependent and given by the equations E_ox = 1.23 − 0.059 pH and E_red = −0.059 pH vs SHE, respectively. In other words, the operating potential ranges of positive and negative electrodes in aqueous based capacitor systems are controlled by the electrolyte pH. For this reason, we have determined the electrochemical performance of the activated carbon DLC Supra 30 (AC) in cells using the various sulfate salt solutions listed in Table I.

Fig. 1 shows the three-electrode cyclic voltammograms (CV) of AC in 1 mol L⁻¹ Li₂SO₄ and 1 mol L⁻¹ BeSO₄. When the scans are realized at potential higher than the water reduction limit (−0.383 V vs SHE in Li₂SO₄ and −0.124 V vs SHE in BeSO₄), the CVs demonstrate only electrical double-layer charging in both electrolytes. By contrast, upon polarization toward potentials lower than the reduction limit at the considered pH, the performance of AC differs significantly in the two electrolytes. A negative current leap related with nascent hydrogen formation and chemisorption in the AC electrode³⁸ starts to appear at potential lower than the limit, and it rapidly gives rise to important oscillations related with di-hydrogen bubbling in case of the cell in BeSO₄, while such clearly visible oscillations are almost not observed with Li₂SO₄. Since the pH of bulk 1 mol L⁻¹ Li₂SO₄ is close to neutrality (pH = 6.5), a low amount of OH⁻ anions produced due to water reduction provokes a sudden pH increase inside the AC porosity to much higher values (after polarization of the cell, the pH measured on electrode surface is higher than 10), and it results in higher over-potential for di-hydrogen evolution. During CV experiments in 1 mol L⁻¹ Li₂SO₄, the onset of oscillations attributed to di-hydrogen evolution starts at about −0.8 V vs SHE. Hence, the negative AC electrode in 1 mol L⁻¹ Li₂SO₄ is capable of i) storing hydrogen produced through water reduction and ii) operating safely up to low potential owing to high local pH in the pores of AC. By contrast, in the acidic 1 mol L⁻¹ BeSO₄ solution (pH = 2.1), the pH within the AC porosity is almost not changed by the produced OH⁻ during water reduction. Consequently, di-hydrogen evolution starts at −0.3 V vs SHE, very close to the thermodynamic reduction potential of water (Fig. 1b), and drives the negative AC electrode to perform in narrow potential range.⁹ The difference of over-potential for AC in the two electrolytes is even better seen during the anodic scan, where hydrogen stored in the AC porosity during the negative scan is oxidized, giving rise to a desorption peak. As it can be seen on Fig. 1, the polarization potential difference between the desorption peak and the thermodynamic water reduction potential is ∼0.78 V and ∼0.42 V in Li₂SO₄ and BeSO₄, respectively. Hence, the desorption activation energy is higher in Li₂SO₄ than in BeSO₄ medium.³⁸

Zoom In Zoom Out Reset image size

The potential extrema of the positive and negative electrodes of AC/AC capacitors in 1 mol L⁻¹ Li₂SO₄ and 1 mol L⁻¹ BeSO₄ with Hg/Hg₂SO₄ reference have been determined by galvanostatic cycling for cell potentials ranging from 0.8 V to 1.6 V (Fig. 2). The positive electrode of the capacitors in 1 mol L⁻¹ Li₂SO₄ and in 1 mol L⁻¹ BeSO₄ operates below the water oxidation limit (marked by the upper dashed line) up to cell potentials of 1.4 V and 1.5 V, respectively. The potential range of the negative AC electrode itself is controlled by the lower dashed-line representing the practical di-hydrogen evolution determined from the previous three-electrode cell experiments. Consequently, in Li₂SO₄ electrolytes, the negative electrode works without any di-hydrogen production in all the considered cell potential range, while in case of BeSO₄ the limit seems to be reached for a cell potential of 1.3 V. Hence, AC/AC capacitors in salt aqueous electrolytes can operate safely up to 1.6 V only when pH is close to neutrality.

Zoom In Zoom Out Reset image size

In order to better evaluate the effect of different hydrogen evolution over-potential in the two electrolytes, a pressure sensor has been connected to AC/AC cells in 1 mol L⁻¹ Li₂SO₄ and 1 mol L⁻¹ BeSO₄. Fig. 3a displays the evolution of gas over-pressure during a potentiostatic floating period of 6 hours at 1.4 V followed by 10 hours at open circuit, and confirms lower gas evolution in case of using Li₂SO₄ as electrolyte. Gas over-pressure in Li₂SO₄ reached 0.29 bar which is nearly half of 0.56 bar in case of BeSO₄. Hence, due to different pH of the electrolytes, the over-potential for di-hydrogen evolution is different, resulting in different amount of di-hydrogen produced at the negative AC electrode. Consequently, the maximum cell potential of AC/AC capacitors with minimum gassing in salt aqueous electrolytes is controlled by the pH of the bulk solution, and reaches its higher value when the pH of the later is close to neutrality, allowing an important pH variation in the pores of carbon when small amounts of OH⁻ ions are produced.

Zoom In Zoom Out Reset image size

As shown in Fig. 3b, the redox reactions related with hydrogen generation at the negative electrode might also affect the self-discharge (cell potential evolution at open circuit) of the AC/AC cells in 1 mol L⁻¹ Li₂SO₄ and 1 mol L⁻¹ BeSO₄. After a potentiostatic period of 30 minutes at 1.2 V, the self-discharge is definitely higher for the cell in BeSO₄. Such high drop of cell potential in case of BeSO₄ electrolyte might be related with the presence of impurities playing the role of shuttles;^39,40 one might anticipate that some ions, in particular Fe²⁺ and Mn²⁺ are extracted from the stainless steel collectors at pH = 2.1.^41,42

CVs of AC/AC capacitors in various sulfate aqueous electrolytes with different pH values are compared in Fig. 4a. The electrolytes with almost neutral pH values (Li₂SO₄, Na₂SO₄, K₂SO₄ and MgSO₄) demonstrate nearly square shaped voltammograms, while in electrolytes with slightly acidic pH (BeSO₄, pH = 2.1 and Al₂(SO₄)₃, pH = 2.9) the CVs exhibit an important current leap at cell potential higher than ∼1.0 V during the positive scan. As discussed previously, such current increase is driven by i) higher hydrogen formation at the negative electrode than in case of neutral electrolytes and ii) surface functionalities formed by oxidation reactions on the positive AC electrode. It is noticeable that the CVs of capacitors with BeSO₄ and Al₂(SO₄)₃ are more deformed below 0.8 V on reverse scan as compared to those with neutral pH electrolytes, revealing more pronounced redox phenomena in slightly acidic pH; the latter are obviously associated with the important current leap observed at high cell potential in these electrolytes.

Zoom In Zoom Out Reset image size

The CVs realized up to 1 V revealed a pure EDL response in all the investigated sulfate electrolytes. As shown in Fig. 4b, the capacitance values obtained from galvanostatic discharge at cell potentials up to 1.0 V are much higher for capacitors with BeSO₄ (and also Al₂(SO₄)₃) than with Li₂SO₄ (and also the other sulfates with pH close to 6.5). The higher capacitance values in BeSO₄ and Al₂(SO₄)₃, despite the lower conductivity of these two electrolytes (Table I), may be related to the small ionic radii of Be²⁺ (0.031 nm) and Al³⁺ (0.050 nm) as compared to Li⁺ (0.060 nm), Na⁺ (0.095 nm), K⁺ (0.133 nm) and Mg²⁺ (0.065 nm). Indeed, assuming that ions are fully desolvated in the pores of AC,¹ the application of formula C = ɛS/d, where d is the distance between the ions and the pore walls, suggests that the double-layer capacitance should be higher for the smallest ions Be²⁺ and Al³⁺ as compared to the other ions.

Hence, from the foregoing, two different behaviors were observed depending on local pH in carbon porosity and related over-potential for di-hydrogen evolution; 1 mol L⁻¹ Li₂SO₄ and 1 mol L⁻¹ BeSO₄ electrolytes are representative of these different redox mechanisms. Fig. 4a shows that in all cell potential range, capacitance is much higher in BeSO₄ than in Li₂SO₄ electrolyte, and it increases with cell potential due to the above afore mentioned pseudo-capacitive contributions.⁵ From 0.8 V to 1.6 V, the capacitance increases from 90 F g⁻¹ to 108 F g⁻¹ in Li₂SO₄ and from 111 F g⁻¹ to 161 F g⁻¹ in BeSO₄. However, when the efficiency (ratio of discharge time/charge time during galvanostatic cycling) is expressed vs cell potential from 0.8 V to 1.6 V (Fig. 4b), one can observe an important drop from 97% to 81% in BeSO₄, while it keeps at around 99% in all cell potential range with Li₂SO₄ electrolyte. The high drop of efficiency in BeSO₄ at cell potential of about 1.3 V reveals that the faradaic processes (hydrogen generation/storage, formation of oxygenated functionalities by oxidation of the positive AC electrode) at the origin of higher capacitance in this electrolyte are highly irreversible. Consequently, the drop of efficiency together with the observed self-discharge in 1 mol L⁻¹ BeSO₄ precludes its application for AC/AC capacitors. Overall, the AC/AC capacitor based on 1 mol L⁻¹ Li₂SO₄ performs up to 1.6 V with far better efficiency as compared to the one with 1 mol L⁻¹ BeSO₄. Data non presented in this report show that the AC/AC capacitors using all the other electrolytes with comparable pH (Na₂SO₄, K₂SO₄, and MgSO₄) demonstrate good efficiency in all cell potential range from 0.8 V to 1.6 V; in case of Al₂(SO₄)₃, the performance is the same as in BeSO₄. Consequently, as far as the highest cell potential is researched, a pH slightly lower than 7 in the bulk electrolyte is a determining criterion for extending the operating cell potential range and for selecting this electrolyte for AC/AC capacitor.

In the previously presented experiments, the electrodes in the form of pressed pellets and sandwiching the separator were placed between two cylindrical current collectors. This kind of assembling differs from the industrial capacitors, where the electrodes are formed by casting the active layer on a metallic foil used as current collector. To further proceed in the optimization of the AC/AC capacitors in salt aqueous electrolyte, coated electrodes with YP 80F activated carbon (further named AC in all this part) have been realized in order to avoid the accumulation of trace amounts of decomposition products at the active material/current collector interface during accelerated ageing by potentiostatic floating.

Fig. 5 represents CVs of cells with stainless steel (Fig. 5a) and nickel collectors (Fig. 5b) in 1 mol L⁻¹ Li₂SO₄ after every 40 hours of potentiostatic floating at 1.6 V. Whatever the collector, stainless steel or nickel, the CVs recorded on the fresh cells (black dashed line) display the typical rectangular shape for an electrical double-layer capacitor. By contrast, the CVs recorded after every 40 hours of floating at 1.6 V reveal a more resistive character for both types of collectors. However, for cells with stainless steel collectors, the capacitive current at cell potential higher than 1 V diminishes with the floating time, while it remains almost constant in case of nickel collectors (Inset in Fig. 5). This decrease of capacitive current reduces the energy density of the device at high cell potential and is most probably related to porosity saturation due to reduction of AC porosity by oxidation or/and corrosion products.^10,11,12,43 After ageing, the presence of corrosion products is revealed by a russet color at the edges of electrodes and current collectors and on the separator; the formation of such products reduces the life-time of the device.

Zoom In Zoom Out Reset image size

After accelerated ageing at 1.6 V, CVs of cells with nickel collectors did not show so pronounced narrowing as in the case of using stainless steel collectors. However, after opening the cell with nickel (−)/nickel (+) configuration, black and pale green residues were found on the separator and at the edges of the positive current collector, and pale green on the negative one. These residues are probably related to the observed pH increase to 7–8 on the surface of the positive electrode and to 10 on the negative one after 120 hours of floating. According to the values of electrodes potentials reached at cell potential of 1.6 V (E₊ = 1.04 V and E₋ = −0.56 V vs SHE), and taking into account the Pourbaix diagram of nickel,⁴⁴ these deposits can be attributed to +III and +IV (black) and +II (pale green) nickel states. However, as shown in Fig. 6 for the case of the capacitor with nickel (−)/nickel (+) collectors, the performance is not affected by the appearance of the deposits, and the cell resistance remains stable till the end of the test. For the capacitor with stainless steel (−)/stainless steel (+) current collectors, a sudden increase of resistance can be noticed after 12 hours of floating at 1.6 V, which probably contributes to the possible perturbation phenomena, i.e., evolution of gases, oxidation of carbon, formation of corrosion products.

Zoom In Zoom Out Reset image size

Galvanostatic cycling realized at 1 A g⁻¹ revealed the same discharge capacitance value of 85 F g⁻¹ for the cells with stainless steel (−)/stainless steel (+) and nickel (−)/nickel (+) collectors. However, the positive electrode of the stainless steel (−)/stainless steel (+) cell displays a capacitance value of 125 F g⁻¹, while the negative electrode of the nickel (−)/nickel (+) cell demonstrates 100 F g⁻¹. Therefore, we decided to build an AC/AC system combining nickel and stainless steel as negative and positive collectors, respectively. Although, this combination did not improve the capacitance of the whole cell, Fig. 6 reveals that this cell presents a lower initial resistance value (0.6 Ω), and the most stable capacitance and resistance values during floating at 1.6 V. Once opening the cell after 120 hours of floating at 1.6 V, there was no corrosion of the positive current collector, although we could still notice green deposits on the negative one.

As shown in Fig. 7 displaying the CVs of both positive and negative electrodes separately, by using a two-electrode cell with reference electrode, the diminished corrosion of the positive stainless steel collector in the nickel (−)/stainless steel (+) system at 1.6 V could be explained by ca. −105 mV shift of the electrode potentials. In the figure, the vertical dashed lines represent the thermodynamic limit for di-oxygen evolution in 1 mol L⁻¹ Li₂SO₄ (pH = 6.5) calculated from the Nernst equation (E_O2 = 1.23 − 0.059 pH = 0.847 V vs SHE) and the experimental limit for di-hydrogen evolution (−0.8 V vs SHE). The value for H₂ evolution was determined by observing the current oscillations on the three-electrode cyclic voltammograms, by analogy with the experiment done in the first part. Considering the nickel(−)nickel (+) system, the formation of deposits found on the separator and current collectors can be explained with help of Fig. 7a. At the negatively polarized electrode, the electrode surface pH increases up to 10, due to the formation of hydroxyl ions (H₂O + e⁻ → H + OH⁻),⁹ and the simultaneously produced nascent hydrogen is either chemisorbed on the carbon surface or combines to form di-hydrogen, provoking oscillations on the CVs (Fig. 7a).⁴⁵ Water reduction results in an electrochemical OH⁻ gradient allowing the chemical precipitation of Ni(OH)₂ (green deposit) at potential E₀ = 0.11 – 0.059 pH⁴⁶ which is in agreement with the appearing current leaps below −0.48 V vs SHE.

Zoom In Zoom Out Reset image size

An increase of the anodic current from 1.0 V vs SHE (Fig. 7a) evidences the oxidation of Ni⁰ to Ni²⁺ or Ni(OH)₂,⁴⁷ and consequently a pale green residue appears on the edges of the positive electrode. Then, in analogy to rechargeable batteries, Ni^II (pale green Ni(OH)₂) may be oxidized to Ni^III (black NiOOH).^48,49 Since NiOOH is a good mixed-conductor, with a high mobility of ionic and electronic species, its presence does not much impede the electrochemical performance of capacitors utilizing nickel collectors.

Considering now the vertical line representative of O₂ evolution in Fig. 7b, it can be seen for the stainless steel (−)/stainless steel (+) system that oxygen evolution may occur at the positive electrode at a cell potential of 1.2 V. However, for this cell, an anodic current leap indicating the electrolyte decomposition, with possible oxidation of the carbon electrode and/or corrosion of the positive current collector starts to be remarkable from a cell potential of 1.5 V. This value is in agreement with our previous studies showing that AC/AC supercapacitors using aqueous Li₂SO₄ with stainless steel current collectors can operate safely up to 1.5 V.¹⁰

Finally, in case of the nickel (−)/stainless steel (+) system operating up to a cell potential of 1.6 V, the anodic current leap attributed to electrolyte decomposition, electrode oxidation and/or formation of corrosion products on the positive current collector is negligible (Fig. 7c). The accumulation of Ni^II and Ni^III derivatives may be responsible for polarizing the surface of electrodes, causing a shift in rest potential (E₀) from 0.285 V to 0.224 V vs SHE.⁵⁰ The potential shift to lower values easily explains the absence of corrosion of the positive stainless steel collector for the cell combining nickel (−) and stainless steel (+). Hence, the application of nickel collectors helps to improve the long time operation of AC/AC supercapacitors based on 1 mol L⁻¹ Li₂SO₄ electrolyte when high cell potentials are applied.

The surface of the activated carbon DLC Supra 30 was chemically modified with phenanthrenequinone moieties by grafting of electroactive 2-amino-9,10-phenanthrenequinone molecules using the reduction of the corresponding cation (see Carbon modification by grafting section), giving the DLC Supra 30-PQ carbon. The mass loading of PQ molecules in the DLC Supra 30-PQ carbon was estimated from TG analysis and cyclic voltammetry²⁰ and is about 20 wt%. At least three measurements were performed on each batch using both techniques (TGA and CV), and the mass loadings were always in the same order of magnitude with an uncertainty of approximately 5% of the mass loading value. Fig. 8 shows the mass loss by thermogravimetric analysis (TGA), together with the CO₂ and CO profiles obtained by thermoprogrammed desorption (TPD) for DLC Supra 30 and DLC Supra 30-PQ. The larger mass loss for DLC Supra 30-PQ, mainly related to CO₂ and CO evolution, indicates an important surface functionalization, which is the proof of quinone grafting. The amount of CO-evolving groups increases from 453 μmol g⁻¹ for DLC Supra 30 to 996 μmol g⁻¹ for DLC Supra 30-PQ. The CO TPD pattern includes contributions from phenols or less stable carbonyl/quinones groups at 675°C.⁵¹ As it is observed on Fig. 8b, the amount of CO₂-evolving groups also increases from 192 μmol g⁻¹ for DLC Supra 30 to 1176 μmol g⁻¹ for DLC Supra 30-PQ, with desorption peaks at 225°C, 475°C and 675°C assigned to strongly acidic carboxylic groups, less acidic carboxylic groups or carboxylic anhydrides and lactones, respectively.⁵² The relatively high value of evolving CO₂ might be related to the presence of free radicals in the reaction medium, causing the formation of surface groups on activated carbon.

Zoom In Zoom Out Reset image size

Fig. 9a shows the nitrogen adsorption isotherms at −196°C for as-received DLC Supra 30 and DLC Supra 30-PQ. Both isotherms display a rounded knee characterizing a wide range of micropore sizes; this is confirmed by the pore size distribution (PSD, Fig. 9b) curves which reveal pores up to ca. 3 nm. After grafting PQ, the BET specific surface area is reduced from 2066 to 649 m² g⁻¹, the micropore volume from 0.908 to 0.282 cm³ g⁻¹ and the mesopore volume from 0.172 to 0.04 cm³ g⁻¹. The interesting feature coming from the PSD comparison of both carbons is a comparable reduction of all pore sizes after grafting of phenanthrenequinone; this is traduced by very comparable values of average micropores size L₀, e.g., 1.16 nm and 1.21 nm for unmodified and grafted carbon, respectively.

Zoom In Zoom Out Reset image size

Whereas literature reports data on electrochemical investigations of quinone-modified activated carbons in acidic and basic aqueous medium,^34,35,53 to the best of our knowledge, data were not reported in media with pH close to neutrality for activated carbon grafted with phenanthrenequinone. Our preliminary studies using a three-electrode cell with DLC Supra 30-PQ as working electrode in 1 mol L⁻¹ Li₂SO₄ have demonstrated a shift of PQ redox potential during the measurements. Since protons are part of the overall redox process, the redox potential is pH dependent, which is particularly pronounced for high porosity carbon powders in aqueous electrolyte. Since the nature of electrolyte and whether the solution is buffered or unbuffered have been found to have a major influence on the electrochemical performance of activated carbon electrodes modified by quinone derivatives,²² we have decided to use buffered solutions as electrolyte in order to eliminate any shift of redox potential during the electrochemical investigations on DLC Supra 30-PQ.

Fig. 10 depicts the cyclic voltammograms of DLC Supra 30 and DLC Supra 30-PQ in biphthalate buffer (pH = 4). The CV of as-received DLC Supra 30 displays a rectangular shaped voltammogram typical of the electrical double-layer (EDL) charging, whereas for DLC Supra 30-PQ, a pseudo-capacitive contribution related to the PQ molecules strongly bonded to the surface of activated carbon takes place together with the EDL formation.³⁵ As a result of grafting PQ molecules, the specific capacitance increases from 82 F g⁻¹ for DLC Supra 30 up to 194 F g⁻¹ for DLC Supra 30-PQ. It is also interesting to note on Fig. 10 that the EDL capacitance of DLC Supra 30-PQ is comparable to that of DLC Supra 30, although the BET specific surface area of the former is 70% lower after grafting. The lowering of S_BET can be explained by the fact that grafting occurs primarily at the entrance of micropores, hindering the access of nitrogen molecules probes to these pores during the gas adsorption measurements.²⁰ However, one has to consider that ions and water molecules in the electrolyte might behave differently from nitrogen. This is in particular revealed by static contact angle measurements on pellet electrodes made from DLC Supra 30 and DLC Supra 30-PQ and showing values of 138.6° and 126.2°, respectively. Hence, due to the polarity of the strongly bonded molecules on the surface of the DLC Supra 30-PQ carbon (and probably also some other surface groups as revealed by the presence of CO₂-evolving groups in TPD), the adhesion energy at the interface between the grafted carbon electrode and the electrolyte could be much larger in comparison to that of unmodified carbon in the same electrolytic medium.⁵⁴ Despite the remarkably lower BET specific surface area of DLC Supra 30-PQ as compared to DLC Supra 30, better wettability of the grafted carbon allows better accessibility of the electrolyte to the pores (even if micropore entrance is slightly closed by quinone molecules), explaining EDL capacitance values in the same order as for the as-received carbon.

Zoom In Zoom Out Reset image size

The effect of electrolyte pH on the three-electrode cell properties of DLC Supra 30-PQ has been compared in 0.1 mol L⁻¹ phosphate (pH = 7.2) and biphthalate (pH = 4.0) buffer solutions (Fig. 11). As expected from literature on grafted graphite powder electrodes,⁵⁵ the anodic and cathodic peaks shifted toward negative potentials by increasing the solution pH; the specific capacitance values were 194 F g⁻¹ to 228 F g⁻¹ in biphthalate buffer and 0.1 M phosphate buffer, respectively. Two redox peaks are observed both in biphthalate buffer and phosphate buffer. Additionally, the reversibility of the redox reaction is smaller in phosphate buffer as compared to the biphthalate one. The peak-to-peak separation between anodic and cathodic peaks (ΔE_p) evaluated from the CVs increases with the electrolyte pH, indicating that the overall process is kinetically much slower in the presence of phosphate buffer.³² At higher pH (phosphate buffer), the peaks current ratio (I_a/I_c) for peaks located at ca. 0.3 and 0 V was found to be greater than unity, which indicates possible kinetic disturbance in the electrode process⁵⁶ contrary to biphthalate buffer.

Zoom In Zoom Out Reset image size

At pH of about 4–10, quinone derivatives undergo a two-electron/two-proton reduction process.⁵⁷ When the protons concentration in the electrolyte is close to the quinone equivalent concentration on the carbon surface, CVs show two waves.³² The wave at more positive potentials is universally interpreted as being due to the 2e⁻, 2H⁺ reaction, while the interpretation of the wave at more negative potential varies. Some authors suggest that it is related with the protons coming from either water or quinones. More recently, Shim and Park^58,59 have argued that it is due to 1e⁻ reduction to form the quinone radical anion. Wang and co-workers⁶⁰ disputed this interpretation suggesting, instead, that it is due to 2e⁻ reduction to form the quinone dianion. Besides, the two peaks in the cyclic voltammogram for both electrolytes could be associated to different types of bonding or different sites of grafting²³ between the phenanthrenequinone molecules and the carbon surface. For example, it has been reported that in case of chemical modification without reducing agent, the derivatization might be occurring through a cationic intermediate. It reacts with surface carbonyl groups of the carbon powder and results in the formation of ester linkages.⁶¹