Articles
Pore Structures and Electric Double Layer Properties of Activated Carbon Derived from Demineralized Spent Coffee Grounds
2021 Volume 89 Issue 6 Pages 573-578
Details
2021 Volume 89 Issue 6 Pages 573-578
To improve the specific surface area (SSA) of steam activated carbon that has been prepared from spent coffee grounds (SCG), a demineralization of the SCG was conducted. As the ash content of the SCG decreased, fine micropores developed during the activation, leading to a significant increase in the SSA. An electric double layer capacitor was assembled and then evaluated, with the evaluation results indicating that the capacitance per electrode weight, the capacitance retention in relation to the current density, and the internal resistance were superior for activated carbon with a higher SSA. However, the capacitance per electrode volume had a maximum value under certain conditions, which were considered well balanced in terms of the SSA and electrode density.
An electric double layer capacitor (EDLC) stores electric energy by utilizing the adsorption-desorption of electrolyte ions on the surface of porous electrodes. Its features include rapid charge-discharge capacity, reduced heat generation, and a long cycle life. Therefore, an EDLC is an ecological electricity storage device that can be applied to energy regeneration.1–3 Until now, the activated carbon that is used for EDLC electrodes has mainly been produced from petroleum-derived resins or imported coconut shells. However, there is now an urgent need to pursue decarbonization by ending the use of fossil resources and utilizing domestic biomass from now on. About 600 thousand tons of spent coffee grounds (SCG) are generated per year in Japan,4 and most of them are incinerated or landfilled.
We have studied SCG and bamboo as possible raw materials for the production of activated carbon and applied them to EDLC electrodes.5,6 However, neither of these types of activated carbon can equal the EDLC performance of activated carbon derived from coconut shells. The specific surface area (SSA) needs to be improved by adopting fine pore formation to achieve a superior EDLC performance. Ash (e.g., alkali metal or alkali earth metal) has the effect of enlarging the pore diameter during activation.7–9 This lowers the SSA and produces a low yield of activated carbon. The ash content of a coconut shell is much lower than that of SCG or bamboo,10 which is why activated carbon derived from coconut shells offers a superior performance. Recently, we have succeeded in increasing the SSA of activated carbon derived from bamboo by conducting a demineralization of the bamboo.11 This type of developed activated carbon mainly consisted of micropores, while the activated carbon derived from non-demineralized bamboo had a micro-mesopore structure. Given that the main component of bamboo ash is potassium, we considered whether the SSA of activated carbon derived from SCG could be improved since its ash also mainly consists of potassium. Our research concept is indicated in Fig. 1. In this study, we prepared activated carbon from demineralized SCG and then evaluated its pore structure and EDLC properties.
Research concept in order to improve the capacitive performance of SCG activated carbon.
SCG that had been discarded from a beverage factory was completely dried at 110 °C. The dried SCG was then sieved to obtain grounds with a particle size of 0.5–2.0 mm. After that, 100 g of the SCG was immersed in 300 ml of ion-exchanged water before being subjected to a boiling-cooling process twice. As a result, the SCG was almost precipitated in the water. After decantation, the solution was stirred for 1 hour following the addition of around 1 L of 8 wt% nitric acid. The SCG was then thoroughly rinsed with ion-exchanged water and dried at 110 °C to obtain demineralized SCG (D-SCG).
2.2 Carbonization and steam activationA crucible containing 70 g of D-SCG was set in an electric furnace. The D-SCG was then heated to 600 °C at a rate of 10 °C/min and carbonized for 1 hour under an N2 stream flowing at 2 L/min to produce D-SCG carbon (D-SCGC). The carbonization yield was calculated using the following equation.
| \begin{align} &\text{Carbonization yield (wt%)} \notag\\ &\quad= \text{Weight of D-SCGC}/\text{Weight of D-SCG} \times 100 \end{align} | (1) |
Next, 2 g of the D-SCGC was activated using the steam activation apparatus reported in our previous paper.12 The D-SCGC was heated to 850 °C at a rate of 10 °C/min under an N2 stream flowing at 200 ml/min and activated by steam that was generated by a saturated steam generator at 80 °C. Several activated carbons with different activation yields were produced by changing the activation time (D-SCGAC). The activation yield was calculated using the following equation.
| \begin{align} &\text{Activation yield (wt%)} \notag\\ &\quad= \text{Weight of D-SCGAC}/\text{Weight of D-SCGC} \times 100 \end{align} | (2) |
For comparison, non-demineralized SCG (N-SCG) was treated as described above to obtain N-SCG carbon (N-SCGC) and N-SCG activated carbon (N-SCGAC).
2.3 Ash analysis of raw material and carbonized sampleTo investigate the composition of the ash contained in N-SCG, D-SCG, N-SCGC and D-SCGC, an elemental analysis was conducted using an X-ray fluorescence spectrometer (model: SEA5120A; manufacturer: Hitachi High-Tech Corporation). The ash content was also measured by burning samples at 850 °C for 1 hour.
2.4 Pore structure analysis of activated carbonN2 adsorption measurements were carried out at the liquid N2 temperature using a gas adsorption instrument (model: BELSORP-max; manufacturer: MicrotracBEL Corp.) to evaluate the SSA and pore size distribution for the activated carbons that had been prepared. The SSA was calculated by means of the BET method.13 The micropore and mesopore size distributions were analyzed by means of the MP method and the BJH method, respectively.14,15
2.5 EDLC assembly and electrochemical measurementThe D-SCGAC was pulverized using a mortar. Electrodes were prepared by mixing 80 wt% activated carbon powder, 10 wt% carbon black, and 10 wt% polytetrafluoroethylene (PTFE) with a small amount of ethanol. This mixture was then rolled into a sheet with a thickness of 100 µm by using a tabletop roll press machine (model: HSMNRP-1; manufacturer: Hohsen Corp.). After it had been stored in a desiccator for at least 12 hours, the sheet was punched into a circular form with a diameter of 15.95 mm (area: approximately 2 cm2) and then heated in a vacuum at 250 °C. After that, the electrodes were removed from the furnace and their weights and thicknesses were measured. The electrodes were then immersed in an electrolyte solution of 1 mol/L triethylmethylammonium tetrafluoroborate (TEMA-BF4)/propylene carbonate (PC) in a glove box, which was evacuated to transfuse the electrolyte into the fine pores of the electrodes. After this impregnation, a two-electrode coin-type cell (model: HS cell; manufacturer: Hohsen Corp.) was assembled.
A constant current (CC) discharge was performed after the completion of a constant current-constant voltage (CC-CV) charge at 30 °C within a voltage range of 0–2.5 V at discharge current densities of 5, 15, 25, 50, 75 and 100 mA/cm2 by using an electrochemical measurement system (model: HZ-7000; manufacturer: HOKUTO DENKO CORP). The capacitance per electrode weight (CW) was determined from the slope of the discharge curve between 1 second after the discharge had started and the moment of full discharge, using the following equation.
| \begin{equation} C_{\text{W}} =- 2I \Delta t/\Delta VW \end{equation} | (3) |
Where I = current (A), t = time (s), V = voltage (V) and W = weight of a single electrode (g).
The internal resistance was calculated from the IR drop at a CC discharge of 50 mA/cm2 after a CC-CV charge of 2.5 V. Impedance measurements within a frequency range of 100 kHz to 10 mHz were conducted under full discharge.
Table 1 shows the results of the ash content measurements. The ash content of the N-SCG was 1.06 wt%, but that of the D-SCG decreased dramatically to 0.07 wt%. The ash content of the N-SCGC condensed up to 4.02 wt% whereas that of the D-SCGC maintained a low value of 0.43 wt%. The inorganic composition obtained from the fluorescent X-ray analysis was reflected in the ash content, as shown in Fig. 2. In this figure, the content rates of Mg, K and Ca are indicated since they have a strong catalytic effect on activation.16 While considerable amounts of Mg, K and Ca were detected in the N-SCGC, almost all of these elements were removed for the D-SCGC.
| Sample | Ash content (wt%) |
|---|---|
| N-SCG | 1.06 |
| D-SCG | 0.07 |
| N-SCGC | 4.02 |
| D-SCGC | 0.43 |
N-SCG: non-demineralized SCG.
D-SCG: demineralized SCG.
N-SCGC: N-SCG carbonized at 600 °C for 1 hour.
D-SCGC: D-SCG carbonized at 600 °C for 1 hour.
Content rates of Mg, K and Ca included in each sample.
The carbonization yields of the D-SCGC and the N-SCGC were 28.9 wt% and 25.6 wt%, respectively, so the value for the D-SCGC exceeded that of the N-SCGC. After the demineralization treatment had been completed for the SCG, the yield of the D-SCG was about 89 wt%, so components other than the ash seemed to have been removed. These components are considered to be low molecular substances that are hard to retain during carbonization. Due to the pre-removal of these components, the carbonization yield of the D-SCGC was a little higher. In addition, as the ash content was low, an effect that inhibits the combustion of carbon is believed to have occurred.16
Figure 3 shows the activation yields in relation to the activation times for the N-SCGAC and the D-SCGAC. The activation yield for the N-SCGAC decreased sharply in line with the activation time until it was almost reduced to ash at 60 minutes. On the other hand, the activation yield for the D-SCGAC experienced a gradual decline, maintaining approximately 15 wt% even when it was subjected to activation for 180 minutes. The reaction rate of the steam activation for the D-SCGC is believed to have slowed due to the ash reduction.
Activation yields in relation to activation times.
Figure 4 shows changes in the SSA in relation to the activation times. The SSA of the N-SCGAC reached a maximum of 1527 m2/g after activation for 50 minutes. SSA measurements were not taken after activation for 60 minutes because of incineration. The SSA of the D-SCGAC increased in line with the activation time, reaching a high value of 2025 m2/g. No papers indicating that SCGAC produced by steam activation can have an SSA of more than 2000 m2/g have been found.
SSA changes in relation to activation times.
The relationship between the total yield and the SSA is shown in Fig. 5. The total yield was calculated by multiplying the carbonization yield by the activation yield and then dividing that value by 100. The numbers given in the graph area are the activation times (unit: min). The plots for the D-SCGAC shifted upwards compared to those for the N-SCGAC, which suggests an improvement in the SSA at the same total yield. The total yield was also improved when the SSA was the same.
The relationship between total yield and SSA. Numbers in the graph are activation times (min).
Figure 6 shows the N2 adsorption-desorption isotherms. The shape of the isotherm for the N-SCGAC was Type IV, for which the adsorption amount increases at both low relative pressure (0–0.1) and medium relative pressure (0.4–0.9), indicating the formation of mesopores in addition to micropores.17 The shape of the isotherm for the D-SCGAC was Type I, for which the adsorption amount rises only at a low relative pressure, so the D-SCGAC was thought to have mainly a micropore structure. However, the slope of the transition curve between low relative pressure and medium relative pressure became more gradual in line with the activation time, indicating an enlargement of the pore diameter.
N2 adsorption-desorption isotherm. Graph legends represent activation times (min). (a): N-SCGAC (b): D-SCGAC
Figure 7 shows the micropore size distribution that was analyzed using the MP method based on the isotherm. Compared to the N-SCGAC, the D-SCGAC had highly developed micropores. Figure 8 shows the mesopore size distribution that was analyzed using the BJH method. The results for the N-SCGAC confirmed that the mesopore size increased in line with the activation progress, with a peak shift from 6.2 nm to 14 nm. On the other hand, there were no notable peaks for the D-SCGAC, except for the maximum size of 2.4 nm. The peak pore sizes seemed to be found only in micropore regions of less than 2.0 nm. Based on the above results, it was confirmed that fine micropores develop in the D-SCGAC while inhibiting pore size enlargement.
Micropore size distribution analyzed using the MP method. Graph legends represent activation times (min). (a): N-SCGAC (b): D-SCGAC
Mesopore size distribution analyzed using the BJH method. Graph legends represent activation times (min). (a): N-SCGAC (b): D-SCGAC
Table 2 provides a summary of the densities of the EDLC electrodes prepared from D-SCGAC. For comparison, YP50F, a commercial activated carbon derived from coconut shell, is indicated. Since the electrode composition was same for each electrode, differences in the electrode densities can be attributed to the activated carbon densities that is inversely proportional to the pore volume. The pore volume of activated carbon is calculated from the N2 adsorption amount at near relative pressure of 1 in Fig. 6. The longer the activation is carried out for, the higher the SSA is but the bulkier the activated carbon becomes, thereby leading to a decline in the electrode density.
| Sample | Weight (mg) | Thickness (µm) | Density (g/cm3) |
|---|---|---|---|
| 80 | 15.10 | 104.5 | 0.72 |
| 100 | 14.35 | 104.0 | 0.69 |
| 120 | 12.85 | 100.5 | 0.64 |
| 140 | 12.20 | 101.5 | 0.60 |
| 160 | 11.15 | 100.5 | 0.55 |
| 180 | 10.55 | 101.5 | 0.52 |
| YP50F | 13.20 | 102.5 | 0.64 |
The name of sample represents activation time.
For comparison, YP50F, a commercial activated carbon derived from coconut shell, is indicated.
Each value is an average of anode and cathode electrode.
Figure 9 shows the capacitance per electrode weight (CW) for each sample. The CW increased in line with the activation time, corresponding to the order of the SSA value. Furthermore, the samples that were activated for no less than 140 minutes retained their CW under high current densities, that is to say, these samples are superior in capacitance retention in relation to current density.
Capacitance per electrode weight CW for each activated carbon prepared from D-SCG. Graph legends represent current density (mA/cm2).
Figure 10 shows the capacitance per electrode volume (CV), which was calculated by multiplying the CW by the electrode density. Unlike the case of the CW, samples with an activation time of 100 minutes exhibited the highest CV until the current density exceeded 50 mA/cm2. Shiraishi et al. reported that CV plots against the BET SSA of activated carbon exhibit a rising convex curve and that there is a maximum value.18 Pore development with activation causes an increase in not only the SSA but also the pore size. As a result, a trade-off relationship exists between the CW and the bulk density. Notably, the results described in this paper also reflect those of the previous report. As the current density rose, the CVs for the samples that were activated for a short time of no more than 120 minutes decreased and samples with an activation time of 140 minutes exhibited the highest CV above 75 mA/cm2. In our previous report, CW of 86.0 F/g and CV of 53.9 F/cm3 were obtained at a current density of 5 mA/cm2 using N-SCGAC activated for 50 min and 40 min, respectively.19 In this study, the highest CW of 92.3 F/g and the highest CV of 57.8 F/cm3 were obtained at a current density of 5 mA/cm2 using D-SCGAC activated for 180 min and 100 min, respectively. These results suggest that demineralization of SCG is effective in improving capacitive performance.
Capacitance per electrode volume CV for each activated carbon prepared from D-SCG. Graph legends represent current density (mA/cm2).
The reason for the reduction in capacitance at a high current density is that electrolyte ions cannot follow the current. Kim et al. described the relationship between the pore structure and the capacitance retention in relation to the current density, concluding that the diffusion resistance, which limits the movement of ions into pores, was the main factor.20 In light of this, we conducted impedance measurements to investigate the diffusion resistance of ions. Figure 11 shows the Nyquist plot obtained from the impedance measurements. The samples that were activated for a longer time had a low diffusion resistance, which appears as a diagonal line with a slope of 45°.21 Ions are considered to have diffused into the pores easily for the long activated samples.
Nyquist plot obtained from the impedance measurements. Graph legends represent activation times (min).
Figure 12 shows the internal resistances calculated from the IR drop on discharge. As the activation time grew longer, the internal resistance dropped. This drop was also affected by the reduction in the diffusion resistance of the ions.
Internal resistance calculated from IR drop on discharge.
To improve the performance of activated carbon derived from SCG, a demineralization of the SCG was conducted. Almost all of the ash contained in the SCG was removed using a nitric acid solution to obtain D-SCG. The D-SCG was then carbonized and activated by steam, after which the activation yield in relation to the activation time experienced a gradual decline, indicating that the activation reaction was slower. Compared to the N-SCGAC, the SSA of the D-SCGAC increased considerably more. This was attributed to the selective development of micropores. For an EDLC using D-SCGAC electrodes, samples with a longer activation time had superior properties in terms of CW, capacitance retention in relation to current density, and internal resistance. When the current density was relatively low, however, the maximum CV could be obtained at the activation time of 100 minutes (activation yield: 34.7 wt%). Under this condition, both the SSA and the electrode density were considered to be well balanced.
Keisuke Kikuchi: Data curation, Methodology, Supervision, Writing – original draft
Keigo Hasumi: Data curation, Funding acquisition, Methodology, Resources, Writing – review & editing
Takashi Fujimura: Funding acquisition, Methodology, Resources, Supervision, Writing – review & editing
K. Kikuchi: ECSJ Active Member
