Waste Biomass-Derived Carbon Anode for Enhanced Lithium Storage

Abstract

Due to increased populations, there is an increased demand for food; thus, battery electrode materials created from waste biomass provide an attractive opportunity. Unfortunately, such batteries rarely sustain capacities comparable to current state-of-the-art technologies. However, an anode synthesized from waste avocado seeds provides high cycling stability over 100 cycles and provides comparable capacity to graphite, around 315 mAh g^–1 at 100 mA g^–1 current density, and readily outperforms graphene in terms of both stability and capacity. This novel electrode provides such capacities as an amorphous carbon without the use of any additives or doped heteroatoms by utilizing capacitance-driven mechanisms to contribute to 54% of its lithium-ion storage. This allows the waste biomass-derived anode to overcome its low apparent diffusion coefficient of 4.38 × 10^–11 cm² s^–1. By creating battery anodes from avocado seeds, waste streams can be redirected into creating valuable, renewable energy storage resources.

1. Introduction

By the year 2050, the human population is projected to exceed 9 billion people,^1,2 requiring food production to increase by an estimated 70%.³ The United States Department of Agriculture (USDA) reported in 2014 that 31% of all food in the United States went unconsumed, ending up in landfills and costing an estimated $160 billion.⁴ Further, waste decomposition in these landfills accounts for 16.4% of the total methane pollution in the United States.⁵ Diversion of food waste from landfills would minimize both revenue loss and environmental stress. An attractive method is to convert unwanted waste into highly valued energy solutions. Biofuel production from biomass has been largely successful throughout both North and South America.⁶ However, biofuels still produce harmful greenhouse gases upon use in combustion engines. Converting waste biomass into battery electrode materials is much more attractive with the current demand in mobile devices and the future demand in electric vehicles (EVs) and renewable grid-scale energy storage applications. A successful waste biomass-derived carbon anode for lithium-ion batteries would not only be environmentally friendly but could also potentially be extremely profitable by valorizing otherwise unused waste.

Ever since the first commercialized lithium-ion battery (LIB) was distributed by Sony in 1991, lithium-ion batteries have used graphite as the anode. Hence, discovering novel carbonaceous materials and utilizing them as battery anodes possess great potential for the discovery of the next generation of LIBs. Much work has already been conducted on various biomass-derived carbon anodes. However, the performance of such anodes varies drastically despite all being synthesized similarly (mainly pyrolysis followed by acid or porogenic treatment), as shown in Table 1. Most studies on novel biomass-derived carbon anodes attribute their respective performances to differences in surface morphology and internal structure without quantifying the specific effects these differences have on their lithium-ion storage mechanisms. Without such studies, trends within the field cannot be generated and optimized, thus stifling the realization of a commercial biomass-derived battery.

Table 1. Summary of Previous Biomass-Derived Carbon Anodes for Lithium-ion Storage.

biomass	synthesis method	reversible capacity	reported number of cycles	rate capability	ref
pomelo peel	pyrolysis, phosphoric acid treatment	181 mAh g–1 at 200 mA g–1	200 (consistent reversible capacity)	stable up to 5 A g–1	(7)
banana peel	pyrolysis, KOH treatment	200 mAh g–1 at 40 mA g–1	10 (inconsistent reversible capacity)	not reported	(8)
coconut oil	incineration, piranha treatment	250 mAh g–1 at 100 mA g–1	90 (consistent reversible capacity)	stable up to 1 A g–1	(9)
green tea leaves	pyrolysis	471 mAh g–1 at 40 mA g–1	50 (consistent reversible capacity)	stable up to 4 A g–1	(10)
coffee oil	dry autoclaving	274 mAh g–1 at 100 mA g–1	250 (consistent reversible capacity)	stable up to 0.5 A g–1	(11)
sugar	pyrolysis, sulfuric acid treatment, “Dewatered” decomposition	209 – 647 mAh g–1 at 20 mA g–1	2 (reversible capacity calculated as average of 2nd charge and discharge cycles)	not reported	(12)
pistachio shell	pyrolysis	130 mAh g–1 at 40 mA g–1	50 (consistent reversible capacity)	stable up to 0.2 A g–1	(13)
peanut shell	pyrolysis, proprietary porogenic agent	1348–1579 mAh g–1 at 40 mA g–1	10 (inconsistent reversible capacity)	not reported	(14)
cotton fiber	annealing, graphene coating	288 mAh g–1 at 50 mA g–1	50 (capacity vs cycle number not reported)	not reported	(15)

Recently, avocados have become one of the most popular foods for the current generation, with over 5 million metric tons produced yearly and no indication of slowing down.¹⁶ Demands for avocado, however, are primarily limited to the flesh of the fruit,¹⁷ which can be eaten raw or processed into products such as guacamole. The rest of the avocado has no current commercial use and is thrown away during processing. As demand continues to grow, avocado farmers and processing plants will produce an abundant supply of avocado seeds, which make up 15% of the avocado’s total weight.¹⁸ To both increase the net worth of avocados and reduce environmental impact from waste, it is beneficial to produce useful, profitable byproducts from avocado seeds. Current research efforts to recycle avocado waste range from the synthesis of activated carbon as an effective adsorbent¹⁹ to the design of large-scale, complex avocado “biorefineries”, which consist of multiple processes for the refinement of both avocado seeds and skin to synthesize a variety of products.¹⁸

In this work, non-graphitic carbonaceous anodes for LIBs were synthesized from avocado seeds, which currently offer little to no economic benefit as a waste byproduct to an already lucrative fruit. Raw avocado seeds were pyrolyzed into a carbonaceous active material without requiring any solvents or catalysts. New to the biomass-derived carbon anode field, the lithium storage mechanism of the novel anode was studied to quantify the contributions to its capacity from both capacitance and intercalation. Using a combination of physical characterizations and electrochemical performance studies, anodes synthesized from avocado seeds prove to be a promising alternative to the current graphite anode, demonstrating superior performance to both commercial graphite and graphene.

2. Results and Discussion

2.1. Material Characterization

Energy-dispersive spectroscopy (EDS) in tandem with scanning electron microscopy (SEM) was used to determine the lack of impurities in AVS, GRT, and GRN samples, as well as their topographies (Figures 1 and S1). The EDS analysis was particularly important for the AVS sample, as the AVS precursor contained impurities of O, Mg, P, K, Ca, and possibly N. The impurity content in the AVS precursor (AV12) was ∼12 wt %, according to its thermogravimetric analysis (TGA) profile (Figure S2). Further, the said impurities were shown to decrease reactivity and adversely affect lithium storage capability (Figure S3); acid purification from AV12 to AVS reduced the impurity content to 1% (Figure S3).

Figure 1.

SEM images of AVS, GRT, and GRN samples.

Raman spectroscopy and X-ray diffraction (XRD) (Figure 2) were utilized to elucidate the internal structure of each material. By measuring the ratio of the disorder-induced band (D-band, ∼1300 cm^–1) to the graphitic band (G-band, ∼1600 cm^–1) in Raman spectra, the degree of amorphousness can be quantified and compared. It was found that AVS (I_D/I_G = 2.59) > GRN (I_D/I_G = 1.10) > GRT (I_D/I_G = 0.74), in order from most amorphous to least amorphous (Figure 2a,b). Biomass-derived carbons are commonly amorphous, as carbon recrystallization for some carbons requires high temperatures >1000 °C for several hours, while other carbons are difficult to recrystallize altogether.²⁰ The XRD pattern in Figure 2c further illustrates the high amorphousness in AVS, as the increase in noise and broadening of peaks demonstrate the presence of turbostratically disordered layers.¹³ Similarities can be seen between GRT and GRN samples in both Raman and XRD plots (Figure 2b,d). Due to GRN being exfoliated sheets of GRT, a similar XRD pattern to and slightly more amorphous nature than GRT is expected.

Figure 2.

Raman spectroscopy and X-ray diffraction signals from AVS (a, c), GRT, and GRN (b, d).

2.2. Electrochemical Performance

Remarkably, the long cycling performance for all three materials (AVS, GRT, GRN) was very similar, delivering similar capacities of ∼320 mAh g^–1 and stable performance over 100 cycles (Figure 3a). This is especially promising for AVS, as biomass-derived anodes commonly deliver a discharge capacity below 300 mAh g^–1 under the same electrochemical reaction conditions.¹¹ Initially, all three batteries exhibit an inflated, gradually decreasing capacity as Li ions are consumed for solid electrolyte interface (SEI) formation. Subsequently, both GRT and GRN demonstrate significant, large “activation” periods in which the capacity performance steadily increases until ∼30 cycles. AVS, on the other hand, demonstrates stable capacity performance immediately after the SEI formation (cycle 3) with comparatively consistent initial capacity retention. This activation period can be attributed to the different rates of electrolyte wetting within the active materials. The significantly longer time taken for the electrolyte to fully penetrate the GRT and GRN electrodes and provide the batteries’ full capacity suggests that AVS has a considerably smaller crystallite size. This is supported by the differences in width between XRD spectra (Figure 2) of AVS and GRT/GRN.

Figure 3.

(a) Cycling performance over 100 cycles at 100 mA g^–1 current density and (b) C-rate performance at steadily increasing current densities (units of mA g^–1), returning to 100 mA g^–1 to demonstrate rate capability.

Although Figure 3a demonstrates comparable charge–discharge performance between all materials at 100 mA g^–1, the rate capability test (Figure 3b) shows key differences between the materials. The rate capability is especially important in the current age of electric vehicles and mobile devices, as fast charging dictates current densities of at least 400 mA g^–1. At this current density, GRN becomes unstable due to drawbacks in the lithium-ion insertion mechanism. However, AVS demonstrates comparably stable performance to commercial graphite for all current densities. Further, AVS delivers 15–30% higher capacity at lower current densities than both GRT and GRN.

Figure 4 illustrates the charge–discharge (C–D) profiles for the tested electrodes. For all three active materials, Coulombic efficiency began at >90% and remained at ∼99.9% for the duration of testing (100 cycles). Distinct differences in the lithium-ion storage mechanism for each material are most apparent when comparing the shape of the curves and contrasting the sloping, gradual curves of AVS to the steep curves of GRT and GRN. While GRT and GRN diffuse Li-ions at primarily <0.25 V vs Li/Li⁺ due to their crystalline layered structure,²⁰ AVS stores Li-ions gradually and consistently over its operating potential. Evidently, the large amorphousness of AVS allows for an even distribution of potentials at which Li-ion insertion is electrochemically favorable. With such a large percentage of lithium ions (>60% of the delivered specific capacity) being stored at a potential <0.25 V vs Li/Li⁺, AVS minimizes any electroplating, acting as an inherently safer anode than both GRT and GRN.²¹

Figure 4.

Charge–discharge plots for cycles 1, 2, 50, and 100 for (a) AVS, (b) GRT, and (c) GRN at 100 mA g^–1. Coulombic efficiencies associated with each cycle are indicated by color-coordinated circles.

2.3. Kinetic Study

A kinetic study was carried out to understand the storage mechanism of lithium ions in AVS. Cyclic voltammetry (CV) was used as the basis of the study, as shown in Figure 5. AVS has similar peaks (between 0.3 V vs Li/Li⁺ and 0.5 V vs Li/Li⁺) to GRT and GRN during the charge process. However, the peaks for AVS are wider and the distance between charge and discharge peak potentials at each sweep rate is greater, especially when compared to GRT. Therefore, although AVS stores lithium using the same mechanism as GRT and GRN—through (de)intercalation to form LiC₆²²—for the majority of its charge–discharge cycles, AVS clearly does not rely solely on (de)intercalation.

Figure 5.

Cyclic voltammetry at varying scan rates for (a) AVS, (b) GRT, and (c) GRN.

By assuming that Li-ion storage can be classified as either capacitance (denoted by k₁) or intercalation (denoted by k₂), contributions from each can be determined from current delivered under particular scan rates (v^1/2)²³

1

For capacitance contributions (k₁), it is worthy to note that pseudocapacitance—the faradic process in which lithium ions are stored directly onto the surface of the carbonaceous electrode—and the double layer effect—the non-Faradaic process in which solvated lithium ions are electrostatically held onto the anode surface—cannot be discerned from one another. However, disordered carbon anodes have mainly exhibited non-faradic reactions, especially for carbon anodes with high specific surface areas.²⁴ Intercalation contributions (k₂) are faradic, with theoretical maximum capacity in graphitic materials as LiC₆ (and thus delivering 372 mAh g^–1).²⁵

Compared to the largely crystalline GRT and GRN, AVS possesses a multitude of active sites for capacitance. As with other amorphous carbons, AVS consists of a disordered bulk of stacks of several layers of graphene organized according to the “house of cards” model.²⁶ Although these stacks can store lithium ions via intercalation akin to crystalline carbon, the interstitial space between the stacks also encourages storage of lithium ions via capacitance. This is reflected in Figure 6, which shows that capacitance contributes to 54% of current induced from the AVS half-cell during cyclic voltammetry studies. Meanwhile, capacitance only contributes to 35% of delivered current in both GRT and GRN half-cells. These could have arisen from defects on the surface of the active material or from slight phase changes induced during slurry creation (under strong mixing with zirconia balls). AVS’s large reliance on capacitance for energy storage explains its sloping charge–discharge profile: unlike GRT and GRN, lithium-ion transfer onto the anode is thermodynamically favored throughout the charge–discharge process in AVS, as the abundance of capacitance active sites provides a mechanism for storage throughout a range of applied potentials.

Figure 6.

Contributions from intercalation and capacitance due to induced current of 0.4 mV s^–1 for (a) AVS, (b) GRT, and (c) GRN. Calculated ratios of capacitance to intercalation are indicated as R.

Lithium-ion diffusion coefficients can be calculated using the Randles–Sevcik equation²⁷ and further analyzing the (de)intercalation processes.

2

The calculated Li-ion diffusion coefficients for the charge process are displayed in Table 2. AVS possesses an apparent Li-ion diffusion coefficient 32% of that of GRT. In addition to its capacitor-like nature, the smaller crystallite size of AVS also explains the difference between GRN and GRT calculated diffusion coefficients. A lower crystallite size, whether measured perpendicular or parallel to the crystalline layers, leads to poor alignment between the layers and explains the apparent, substantial resistance to lithium-ion transport within the layers in AVS.²⁸ However, the AVS lithium-ion capacity is still equivalent to that of GRT and GRN despite the higher resistance due to its utilization of capacitance-driven mechanisms (which exist outside of the layers), as discussed earlier.

Table 2. Apparent Diffusion Coefficient for the Charge Process, Calculated from Cyclic Voltammetry.

material	diffusion coefficient (×1011 cm2 s–1)
AVS	4.3775
GRT	13.749
GRN	8.1202

Further studies should be conducted to elucidate the internal structure of AVS and to determine the lithium storage mechanisms of other biomass-derived carbon anodes. Such studies will determine the key characteristics that make certain biomass-derived anodes perform far superior to others. These results can help to both further optimize existing successful biomass-derived carbons (such as AVS) and inform future studies on new biomass-derived carbon anodes.

3. Experimental Methods

3.1. Synthesis of Waste Biomass-Derived Anodes (AVS)

Hass avocados were pitted, and the seeds were rinsed with water to remove remnant avocado flesh. Seeds were chopped to a grain size no. 10 (around 2 mm). Avocado seed pieces were then pyrolyzed in a tubular furnace under an argon atmosphere from room temperature to 750 °C at a heating rate of 10 °C min^–1. Furnace temperature remained at 750 °C for 2 h, which was then allowed to return to room temperature. The carbon yielded ∼50% of the original avocado seed mass. The carbon pieces were ground into fine particles and then purified with a 1:1 vol % solution of H₂SO₄/H₂O by overnight stir-mixing at room temperature. The acid was removed from the carbon powder by three cycles of washing with excess water and subsequent drying. The treatment yielded ∼90% mass of the untreated powder, representing a ∼45% total mass yield from the original raw avocado seed.

3.2. Characterization of Active Materials

Crystallographic information of active materials was gathered using powder X-ray diffraction (XRD) using a Rigaku diffractometer (SmartLab) in Bragg–Brentano mode. Raman spectroscopy of the powders was performed with a Thermo Fisher Scientific DXRTM Raman microscope (red, 633 nm, 5 mW). Scanning electron microscopy (SEM) using a JEOL NeoScope JCM6000 Benchtop (secondary electron mode) was used to compare the topography of all active materials. The same machine was used for energy dispersive spectroscopy (EDS) to obtain the chemical composition of the samples. Thermogravimetric analysis (TGA) was conducted in a TA Instruments TGA i1000 under an air atmosphere from room temperature to 1000 °C at a heating rate of 10 °C min^–1.

3.3. Lithium-Ion Half-Cell Battery Construction

Active materials—AVS, as prepared above; nano-plated graphene, Sigma-Aldrich (GRN); graphite, MTI (GRT)—were prepared with corresponding compositions of 8:1:1 mass ratios of active material/Super P carbon/poly(vinylidene fluoride) (PVdF) binder. N-Methyl-2-pyrrolidone (NMP) solvent was added until the desired slurry consistency was achieved. Slurries were mixed in a THINKY Mixer for 30 min. Slurries were then cast onto copper foil using the doctor blade method and dried overnight at 80 °C under vacuum to remove excess NMP. Then, 12 mm diameter disks were punched out from the dried casting and used as the negative electrodes in CR2032 coin cells with lithium foil as the counter electrode. A 1.0 M LiPF₆ in DEC:EC (with a volume ratio 1:1) solution was used as the electrolyte, and Celgard 2500 was used as the separator in the lithium-ion half-cells.

3.4. Electrochemical Performance Studies

Batteries made from the different active materials were subjected to both galvanostatic and potentiostatic studies. Before each study, batteries were allowed to rest for 24 h and cycled for three cycles at an applied current density of 25 mA g^–1 for the stable formation of SEI. Cycling studies were conducted in an Arbin cycler for 100 cycles by subjecting half-cells to 100 mA g^–1 current density according to the mass of the active material in each battery. Rate studies were conducted by cycling at 25, 50, 100, 200, and 400 mA g^–1 for 10 cycles each and then returning to 100 mA g^–1 for 10 additional cycles to demonstrate rate capability. Cyclic voltammetry (CV) was conducted in a Gamry potentiostat/galvanostat by cycling batteries from 0.01 to 2.0 V vs Li/Li⁺ under 0.2, 0.4, 0.8, and 1.6 mV s^–1 scan rates for two cycles. Only data from cycle 2 is presented, as reactions corresponding to SEI formation dominate the CV plots from cycle 1.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.0c02389.

• EDS comparison of all materials and the AVS precursor (AV12); TGA comparison of AVS and the AVS precursor (AV12); cycle performance of AVS and the AVS precursor (AV12) (PDF)

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

This work was supported in part by CONACYT-SENER project no. 274314 and NSE-CBET award #1804300.

The authors declare no competing financial interest.