CHAPTER 1: Engineering the Architecture of 3D Graphene-based Macrostructures

Carbon is one of the most abundant elements on Earth with unique mechanical, thermal and electrical properties. Carbon has raised a lot of interest in the scientific community because of its intriguing properties and structural variability. The possibility of forming three different strong covalent bonds (with sp¹, sp² and sp³ hybridization) makes carbon a very promising element for both material scientists and engineers.

Carbon can be found in different allotropic forms depending on its hybridization and crystalline structure (Figure 1.1). For instance, the first non-amorphous allotropic form of carbon discovered was graphite. In graphite, the carbon atoms are arranged in a hexagonal structure through sp² hybridization (Figure 1.1b). In diamond, the carbon atoms have an sp³ hybridization forming four tetrahedral bonds with the four nearest neighbors to create diamond cubic unit cells (Figure 1.1c). In 1985, fullerenes or buckyballs were discovered by Kroto et al. as an unexpected result of investigating particles found in space.¹  The 0D fullerene carbon allotrope forms a hollow cage of carbon atoms connected by single and double bonds, resulting in twelve pentagons, where each pentagon is surrounded by five hexagons (Figure 1.1d).² 

Soon after the discovery of fullerenes, carbon nanotubes (CNTs)^3–5  and carbon nano-onions (CNOs)⁶  were reported in 1991 and 1992, respectively (Figure 1.1e and f). CNOs consist of multi-layered spherical or polyhedral shaped closed carbon shells with a structure resembling that of an onion (Figure 1.1f). The 1D CNTs consist of a tube with diameters generally in the nanometer range and are broadly distinguished by the number of concentric walls that make up the tube (Figure 1.1e). Generally, there are two different types of CNTs, single wall carbon nanotubes (SWCNTs) and multi-wall carbon nanotubes (MWCNTs).

Unrolling a SWCNT results in the last carbon allotrope discovered: graphene (Figure 1.1g). The so-called 2D graphene describes a single layer of carbon taken from a 3D graphite block, first observed using electron microscopy in 1962.⁷  Forty-two years later graphene was “rediscovered”, isolated, and characterized by Novoselov et al.⁸  Graphene has a special set of properties, such as high electrical and thermal conductivity,⁹  enormous specific surface area (1168 m² g⁻¹),⁹  and larger nonlinear diamagnetism than graphite,¹⁰  which set it apart from other allotropes of carbon. It is approximately 100 times stronger than steel, yet with a much lower density, having one of the largest strength-to-weight ratios observed.¹¹ 

Given these impressive properties, graphene and graphene-based materials have great potential in numerous applications¹²  such as energy storage,^13–17  nanoelectronics,^18,19  sensors,^20,21  catalysis,^22,23  and composites,^24–26  among others.²⁷  However, in many cases we do not observe the full theoretical potential of the synthesized composites.²⁸  These sub-par outcomes have been attributed to poor graphene dispersion due to π–π stacking interactions between several graphene sheets. One of the best strategies to overcome poor graphene dispersion, while keeping the intrinsic properties of graphene, is breaking the symmetry of the single sheets by introducing oxygen functional groups in the backbone. In this context, graphene oxide (GO) is widely used as a precursor of graphene composites because it is economical to fabricate on a large scale and easy to process.²⁹ 

Today GO is produced in large quantities through chemical exfoliation of graphite, known as the Hummers' method.^30,31  The technique is based on the principle of oxidizing graphite by treatment in a mixture of strong acids (e.g. H₂SO₄) and oxidizing agents (e.g., NaNO₂, KMnO₄) for 2 hours at 45 °C. Then the reaction mixture is washed in an ice-bath with H₂O₂ to remove the residual KMnO₄. This process introduces several oxygen functional groups such as epoxides, hydroxyl groups, and carboxylic acid, among others (Figure 1.2),³²  making GO hydrophilic and easily dispersible in water via ultra-sonication. The oxidized carbons (i.e., sp³ carbons) and lattice defects of GO can be later eliminated by a reduction step (e.g., chemical, or thermal) to recover the graphene-like properties (i.e., sp² carbons).

However, like any other carbon allotrope, graphene as a bulk material has a strong propensity to form irretrievable agglomerates due to strong π–π interactions among individual graphene sheets. This leads to inadequate exploitation of isolated graphene layers for practical applications. In order to overcome this problem, the integration of 2D graphene nanosheets into 3D macrostructures, and ultimately into a functional system, such as aerogels has been recognized as a progressively critical approach during the past five years. Their intensively interconnected networks, enormous surface area, intense porosity, remarkable sturdiness, and superior graphene building blocks endow a plethora of exciting features that make them extremely suitable for a broad range of clean energy and environmental applications. However, in order to truly harness their potential, one must understand how the design and assembly of these 3D graphene networks impact their final properties. In this chapter, we explore the various types of graphene-based aerogels reported to date and how their architecture impacts their ultimate performance.

Graphene aerogels (GA) were first synthesized in 2010 by two independent groups, Xu et al.³³  and Worsley et al.³⁴  Both groups used GO as the precursor but different gelation processes (e.g., covalent vs. noncovalent). Here we describe the different methods to synthesize GAs.

Xu and co-workers reported a hydrothermal route involving a 180 °C treatment in a pressure vessel for 12 hours to simultaneously reduce and gel an aqueous GO suspension.³³  For the hydrothermal route, gelation occurs upon reduction because the electrostatic repulsion is eliminated due to loss of oxygen functionalities in GO. Local regions on the GO sheets then become hydrophobic and are prone to noncovalent bonding (e.g., π–π stacking) with reduced regions on nearby sheets forming a physically crosslinked gel. The properties of the aerogel achieved by freeze-drying were highly correlated to the starting GO concentration of the suspension. If the concentration was below 1 mg mL⁻¹, no gel was formed. The duration of the hydrothermal treatment also proved to be significant in determining the properties of the aerogel. Both the maximum pressure and treatment time determined the degree of reduction, which impacted density, conductivity, and crosslinking (Figure 1.3).³³ 

Following this seminal report, which demonstrated the properties of graphene sheets in an aerogel form, a number of related studies were inspired.^35,36  Tang et al. reported a noble metal promoted self-assembly of GO gels.³⁷  The method entailed using glucose to reduce a metal salt (e.g., chlorides of Au, Ag, Pd, Ir, Rh, or Pt, etc.) dissolved in an aqueous GO suspension, to metal nanoparticles which created strong cross-links between the GO sheets. The essential role of the noble metal nanoparticles was evidenced by dissolving the metal in aqua regia, which led to the degradation of the aerogel. High electrical conductivities and good mechanical strength were also reported.³⁷ 

Xu et al. reported the use of DNA to cross-link GO sheets. In this case, a solution of double-stranded DNA (dsDNA) was added to a GO suspension and heated to 90 °C for 5 minutes.³⁸  The elevated temperature lead to an unwinding of the dsDNA to single-stranded DNA (ssDNA) chains, which made noncovalent bonds between GO sheets. Despite relying primarily on physical crosslinking, these GO/DNA gels showed remarkable chemical resistance and mechanical strength. Shi et al. used glutathione to simultaneously serve as the cross-linker and reducing agent, yielding GAs that were doped with nitrogen and sulfur.³⁹ 

In addition to initiating gelation via noncovalent and chemical cross-linkers, many researchers have leveraged the abundant chemical functionality native to GO to induce self-assembly of GO suspensions.^40,41  Strong bases, such as ammonium hydroxide, can induce self-assembly of GO suspensions with crosslinking analogous to that found in resorcinol-formaldehyde (RF) sol–gel chemistry. Nuclear magnetic resonance (NMR) spectroscopy techniques reveal the appearance of sp³ carbon as well as –CH₂– and –CH₂O– cross-linkers after gelation of the GO suspension in the presence of ammonium hydroxide at 85 °C, indicating some covalent bonding between GO sheets during gelation (Figure 1.4).⁴² 

Further, a number of chemical reducing agents can aid in the self-assembly process similar to that observed using hydrothermal treatment by Xu et al.³³  Broadly speaking, the self-assembly mechanism consists of the clustering of partially reduced GO sheets as their hydrophilicity decreases. The sheets assemble randomly as the reduced GO clusters are formed and water is excluded from the hydrophobic reduced GO gel, resulting in volume shrinkage. A number of chemical reagents, such as NaHSO₃, Na₂S, ethylenediamine, ammonia, hydroiodic acid, and hydroquinone, have been used to drive GO gelation.^43,44  As many chemical reducing agents are hazardous to chemical workers and/or the environment, there have been several studies focused on “green” reducing agents, such as ascorbic acid. Zhang et al.,⁴⁵  and many other researchers, have used ascorbic acid (i.e., vitamin C) to initiate GO gelation,^46–48  resulting in well-formed aerogels with improvements in electrical and mechanical properties. In addition to being non-hazardous, ascorbic acid is also a mild reductant, and as such no gaseous products are evolved (which tend to disrupt or completely destroy the integrity of the gel). Ji et al. utilized carbohydrates as both reductant and morphology orienting agents in GA synthesis.³⁵  Alternate “green” reductants that have also been reported include tannic acid, dopamine, and amino acids.^49–52 

In another work, GAs with densities of less than 3 mg cm⁻³ were assembled via a one-pot method at the oil-water interface of a GO emulsion (Figure 1.5).⁵³  This emulsion was prepared using a cyclohexane/water mixture in the presence of sodium bisulfite under ultrasonication. The GO emulsion gelled at 70 °C over 12 hours via the gradual removal of the oxygen functionalities. Here, sodium bisulfite served as both reductant and a co-emulsifier due to the salt effect.⁵⁴  Using this method, a cellular pore morphology was formed, which enhanced the mechanical robustness of the aerogel. Finally, gamma ray irradiation was used by He et al. for self-assembly of porous honeycomb GAs.⁵⁵ 

Thermal treatment, chemical reagents, and hydrothermal reduction in an autoclave can be used to reduce GO aerogels to GAs.⁵⁶  Hydrazine, borohyrides, aluminum hydrides, and hydrohalic acids are the most commonly used chemical reagents. Sudeep et al. reported a controlled reduction process to reduce GO gel covalently bonded with resorcinol-gluteraldehyde using hydrazine monohydrate vapor at 50 °C under vacuum for 12 hours.⁵⁷  The reduced GO had an electrical conductivity of 3.4 S m⁻¹ and exhibited good adsorption capacity for CO₂ storage. Tang and co-workers used magnesium vapor to reduce GO aerogels.⁵⁸  The freeze-dried GO aerogel was heated in an ampoule with magnesium powder at 700 °C for 5 hours. After the magnesiothermic reaction, the reduced GO was decorated with MgO nanoparticles, washed with acid, and freeze-dried again. The final GA retained the original morphology with densities as low as 1.1 mg cm⁻³ and had an electrical conductivity of 27 S m⁻¹.⁵⁸  Mi et al. reported 3D highly compressible, elastic, anisotropic, cellulose/graphene aerogels (CGAs) prepared by bidirectional freeze-drying (Figure 1.6).⁵⁹  When the GO content was further increased to 40%, both pore size and aspect ratio decreased, which might be due to the nucleation effect of GO dominating and inducing the formation of smaller ice crystals. The bidirectionally aligned porous structure gave the as-prepared GA outstanding compression properties and was able to recover 99.8% and 96.3% when compressed to 60% and 90% strain, respectively. The combined physical properties of a low density of 5.9 mg cm⁻³ and a high surface area of 47.3 m² g⁻¹ synergistically led to a remarkable absorption capacity of 80–197 times of its own weight.

The other common reduction method is thermal treatment in inert gas. Thermal annealing is one of the most effective methods of achieving high electrical conductivity in the GA. Annealing at 800–1100 °C under inert gas produces aerogels with conductivities of ∼100 S m⁻¹.³⁴  Nonetheless, when using even higher thermal annealing temperatures (1500 to 2500 °C), additional improvements in the crystallinity of the graphene sheets are realized. These improvements are clearly reflected in the Raman spectra, oxidative thermal stability, electrical conductivity, and mechanical properties of these GAs (Figure 1.7).^60,61 

For example, the electrical conductivity can be increased 5–6 times the values recorded at lower temperature anneals. This work shows the critical role of crystallinity in determining the physicochemical properties of GA. Thermal reduction in a furnace is most common, but in some work other means are used. For instance, Hu and co-workers reported the synthesis of ultralight GAs by microwave irradiation. Those aerogels showed densities as low as 3 mg cm⁻³ and yet the structure fully recovered without any fracture after 90% compression.⁶² 

Templating is a process where an ordered or relevant structure (i.e., the template) having a length scale of micro- or nanometers is filled with another material and the template is subsequently removed, thereby leaving an imprint of the template on the filled material.⁶³  In the field of aerogels, templating is mainly used to control the pore size distribution and morphology as these two parameters determine the physical properties of an aerogel. The templating method is classified into two types; (1.2.2.2) hard templating, which uses a rigid material with a stable structure,^64,65  and (1.2.2.1) soft templating, which is a cooperative self-assembly process based on inter- and intramolecular interactions between the surfactant and guest species.⁶⁶ 

Mesoporous carbon microspheres/graphene composites (MCMG) were synthesized in situ via a soft template method by Chen et al. using cetyltrimethylammonium bromide (CTAB) as the structure-directing agent, and an aqueous mesophase pitch (AMP) and GO as the carbon sources.⁶⁷  The authors claimed that the negatively charged GO becomes positively charged upon addition of CTAB in the pH range 7–13, and observed a strong electrostatic interaction between the GO/CTAB composites and the AMP molecules at pH 12. Subsequently, the CTAB/AMP micelles with a spherical shape were formed between the GO sheets through self-assembly.⁶⁷ 

An interface-induced co-assembly process was adapted by Liu and co-workers to fabricate a composite of ordered mesoporous carbon/graphene aerogel (OMC/GA).⁶⁸  They employed a strategy where GA acts as a macroporous substrate and a triblock copolymer F127 as a soft template, where resol is the carbon source. The macroporous graphene network was covered by highly ordered mesoporous carbon of 9.6 nm diameter and the orientation of the mesopores was tuned by varying the ratio of the components. The resol-F127 monomicelles, when mixed with GA, gradually deposited on the macropore walls of GA via non‐covalent interactions, such as hydrogen‐bonds, amphiphilic interactions, and π–π interactions, to obtain such ordered structures.⁶⁸  An all‐solid‐state supercapacitor (ASSS) based on OMC/GA with vertical mesopores exhibited an outstanding specific capacitance (44.3 F g⁻¹) at 5 mV s⁻¹, and high power density with fast charge/discharge rate (≈3545 W kg⁻¹ in less than 3.6 s).

Zhang et al. proposed a versatile technique based on soft bubble templating and freezing to fabricate 3D bubble‐derived graphene foams (BGFs) and 2D bubble‐derived graphene porous membranes (BGPMs).⁶⁹  This technique can be extended to assemble other nanomaterials as building blocks into macroscopic configurations. Hierarchical structures of well‐aligned macroscopic spherical pores were formed by templating of bubble clusters, and random minor pores from ice templating. The volume ratio of bubble clusters to the GO dispersion, the concentration of GO dispersion, freezing rate, and size of bubbles controlled the final architecture of the graphene sheets. The authors proved that the optimized volume ratio to prepare a stable mixture is 1 : 1, and the concentration of GO dispersion needs to be higher than 5 mg mL⁻¹. Flexible sensors made with BGF/polydimethylsiloxane (PDMS) composite exhibited excellent resistance change to a compressive strain of 30%.⁶⁹ 

An emulsion templating technique was employed to synthesize porous materials by forming an emulsion of GO containing hexane droplets by Li et al.⁷⁰  Hexane droplets with diameters in the range of several tens of micrometers to about 200 µm were dispersed homogeneously in a GO dispersion.

Barg and co-workers reported a novel self-assembly strategy for the fabrication of chemically modified graphene cellular networks (CMG-CNs) via a multi-step soft/hard template mechanism that combined emulsion and ice templating.⁷¹  GO acts as a surface-active amphiphile, self-assembling at the interface between the oil droplets and the water phase, and stabilizing the GO emulsion for several months (Figure 1.8).

Zhang et al. reported a double-layer templated graphene (DTG) with two non-stacked graphene layers, separated by numerous mesosized protuberances extending from the graphene layers. MgAl-layered double oxides (MgAl-LDO) were used as templates for the chemical vapor deposition (CVD) – a mediated synthesis of the novel DTG materials.⁷²  After the deposition of the graphene layer, the products were purified through hydrothermal reactions with sodium hydroxide and hydrochloric acid. The samples were then filtered, washed, and freeze-dried to yield a 3D porous double-layer graphene. The authors also reported that the defect-rich DTG samples were hard carbons [i.e., not graphitizable and exhibited high ORR current even after heat treatment at 1600 °C (Figure 1.9)].

Another common technique to obtain 3D graphene macroscopic structures with a foam-like network, i.e., graphene foam (GF), is template-directed CVD.⁷³  A porous 3D interconnected nickel foam is used as a scaffold and methane is introduced as a carbon source which decomposes at 1000 °C under ambient pressure. Graphene films later precipitate on the surface of nickel foam. Due to the difference in thermal expansion coefficient, the films have ripples and wrinkles. A polymethyl methacrylate (PMMA) layer is coated on the graphene-Ni foam before etching away the Ni to avoid the collapse of the graphene network during the etching process. The Ni template is removed by thermo-chemical etching. The PMMA layer is later dissolved in hot acetone to yield a monolith of continuous and interconnected 3D graphene networks (Figure 1.10).

The free-standing graphene foam has a density of 5 mg cm⁻³ which corresponds to 99.7% porosity with a surface area of 850 m² g⁻¹ and when infiltrated with polydimethylsiloxane (PDMS), the electrical conductivity of this foam/composite is as high as 10 S cm⁻¹ for 0.5 wt% loading. Hence, the PDMS/GF composites can be used as stretchable conductors due to their excellent electromechanical stability.⁷⁴  The use of polystyrene (PS) beads or spheres as hard sacrificial templates to create an ordered porous architecture in graphene foams is also common. PS latex spheres of 280 nm diameter were assembled with GO to build up a sandwich-type composite film, followed by heat removal and simultaneous reduction of GO. The 3D GF exhibited a high specific surface area of 402.5 m² g⁻¹. PS microspheres were uniformly wrapped by crinkled GO sheets due to electrostatic interaction. The highly oriented laminar and macroporous structure of the free-standing GF was preserved after removal of the template via calcination at 800 °C (Figure 1.11).⁷⁴ 

Fang and co-workers reported a low-cost iron oxide hard template strategy to create highly wrinkled graphene film (HWGF) with a hierarchical pore structure. The hierarchical porosity and high packing density were achieved by capillary compression in the presence of Fe₃O₄ nanoparticles and the generated HWGF exhibited a surface area of 383 m² g⁻¹ along with a high capacitance of 242 F g⁻¹ at low current densities (Figure 1.12).⁷⁵  Fe₃O₄ nanoparticles (NPs) were homogeneously embedded between the stacked graphene nanosheets, which is attributed to steric hindrance effect, endowing graphene sheets with highly wrinkled morphology.⁷⁵ 

In addition, commercially available polymer foams (such as melamine foam) have been used as supporting frameworks to fabricate graphene oxide/graphite nano-platelets GO/GNP composite aerogels with conductive channels. However, unlike other techniques, the melamine foam framework was carbonized to form conductive networks with a homogeneous covering of reduced GO/GNP sheets.⁷⁶ 

Powder metallurgy templating combined with CVD annealing is another route that has been used by Wang et al. to prepare free-standing nitrogen-doped graphene foams (NGFs).⁷⁷  The authors used melamine as a precursor, which acted both as a carbon and nitrogen source. Melamine and Ni powders were evenly mixed, ground, and pressed. Here, the Ni powder served as both a template and as a catalyst to form the 3D porous structure. The pellets were then annealed in a CVD furnace with H₂/Ar flowing under negative pressure. During annealing, the melamine decomposed into carbon and nitrogen atoms which permeated into the Ni particles and deposited on the surface to form a 3D network. The Ni template was later removed by pickling, cleaning, and drying the structure to obtain a self-supported N-doped graphene foam.⁷⁷ 

A continuous microporous 3D GF was synthesized by Lu et al. by means of combining porous metals through the reduction of metallic salts and CVD (Figure 1.13).⁷⁸  For the synthesis of metallic salts, iron and nickel chloride were used as precursors, pressed into pellets and subjected to hydrogen reduction in a one-step CVD process at a temperature range of 600–1000 °C. The authors found that by increasing the reduction temperature from 600 °C to 900 °C, the thickness of Ni ligaments increased from 0.5 µm to 3 µm, and the size of pores was in the range of 0.5 to 5 µm. Further increasing the temperature to 1000 °C, methane or other hydrocarbon gases were introduced as the carbon source. The as-generated carbon atoms diffused and dissolved into the Ni ligaments because of the high solubility of carbon in Ni. Upon cooling, the dissolved carbon atoms segregated and precipitated onto the surface of Ni ligaments, which was followed by graphene nucleation and propagation over the Ni ligaments.⁷⁸ 

The physicochemical properties of GAs are strongly dependent on how they are assembled. Mechanical properties can vary widely depending on whether the cross-links between the sheets are physical or chemical. Electrical properties depend on low resistance connections between graphene sheets. The introduction of electrochemically active or catalytic molecules can add functionality. The following section will explore how incorporating different elements into the carbon matrix of GAs can impact its structure and functionality.

One of the most important, inexpensive, and useful synthetic methods to obtain GAs was reported in 1989 and is based on the resorcinol-formaldehyde (RF) method for producing carbon aerogels (CAs).⁷⁹  In 2011, Worsley et al. synthesized 3D graphene assemblies by adding different concentrations of RF reactants to a 10 mg mL⁻¹ GO suspension.²⁹  The RF units preferentially nucleated and grew on the surface of GO sheets, covalently bonding them together. A lower RF content produced a graphene assembly with a higher sp² carbon content and higher degree of exfoliation contrary to the gels with a higher RF content. Changing the synthetic parameters and RF contents resulted in a wide range of surface areas (600 to 1200 m² g⁻¹), pore volume, and pore size.⁸⁰ 

Lim et al. drastically reduced the gelation time of GA from several days to 1–2 hours by crosslinking RF and GO using hydrochloric acid (HCl) as a catalyst and acetonitrile as the solvent.^81,82  The GO was suspended in acetonitrile instead of water, followed by the addition of RF and HCl. The GO-RF gels were supercritically dried and carbonized at 1000 °C to obtain a GA with similar surface area and porosity of the RF-derived CA (Figure 1.14).⁸¹ 

Recently, Scaffaro et al. synthesized an ultralight graphene-based aerogel (GPA) by coupling GO and an amino terminated polyethylene glycol (PEGNH₂) by carbodiimide (EDC) in an aqueous environment, followed by freeze-drying (Figure 1.15).⁸³  The GPA showed an ultralight and highly porous (99.7%) network with good mechanical properties. Furthermore, cytocompatibility and hemolysis assays of the GPA exhibited no toxicity in vitro at the tested doses.⁸³ 

A multimodal pore graphene/carbon aerogel was reported by Zhang et al.⁸⁴  The hierarchical aerogel was synthesized via one-step carbonization of graphene crosslinked polyimide (PI) aerogel, avoiding the use of harmful formaldehyde (Figure 1.16). The incorporation of graphene sheets into carbon aerogels reduced the pore size while increasing the amount of micro- and mesopores. The as-prepared graphene/carbon aerogel showed a high specific surface area of 998.7 m² g⁻¹ and specific capacitance of 178.1 F g⁻¹ in 6 M KOH at a current density of 1 A g⁻¹, which is much higher than that of pure carbon aerogels (193.6 m² g⁻¹ and 104.2 F g⁻¹).⁸⁴ 

To explore new functionalities of GAs, such as electrocatalysis or electrode fabrication, researchers have investigated the use of transition metal ions for 3D graphene assembly.^85,86  Chen et al. synthesized a graphene/CeO₂ aerogel using a one-step in situ electrochemical method.⁸⁷  The MDGA was synthesized by freeze-drying a graphene/CeO₂ colloidal solution, which in turn was obtained via electrochemical exfoliation of a graphite anode and in situ deposition of CeO₂ nanoparticles on the resulting graphene sheets using ammonium sulfate and cerium nitrate salts as electrolytes (Figure 1.17). An increase in the concentration of cerium salts in the electrolyte enhanced the Faradaic reactivity of the graphene/CeO₂ hybrid aerogels.⁸⁷ 

Wei et al. prepared Ni-doped graphene/carbon cryogels (NGCCs) using a Ni²⁺ catalyst and adding resorcinol and formaldehyde (RF) to a GO suspension.⁸⁸  The Ni²⁺ catalyst improved the crosslinking between GO and RF, strengthening the cryogel. Freeze-drying and carbonization under an inert atmosphere yielded the Ni-doped aerogel. Ni²⁺ ions were reduced to Ni particles during the carbonization process and thus embedded in the interconnected structures.⁸⁸ 

Molybdenum disulfide (MoS₂) has been used for the hydrothermal synthesis of MoS₂-GA hybrids. Hou et al. prepared a MoS₂/nitrogen-doped GA to study the application of the aerogel as a catalyst for hydrogen evolution in microbial electrolysis cells. The authors observed a significantly higher hydrogen production rate (0.19 m³ H₂ m⁻³ d⁻¹) compared to pristine MoS₂ nanosheets and N-GAs.⁸⁹  Worsley et al. infiltrated a GA with ammonium thiomolybdate (ATM), which upon thermal reduction resulted in MoS₂ sheets layered on graphene sheets in the GA. This MoS₂-GA hybrid exhibited a very large surface area (ca. 700 m² g⁻¹) and retained the native conductivity of the GA (1.12 S cm⁻¹). With 50 wt% MoS₂, the aerogel proved to be an efficient hydrogen evolution catalyst with a low overpotential.⁹⁰  Zhang et al. combined defect-rich MoS₂ nanosheets and conductive graphene nanosheets (GNS) to obtain a hybrid MoS₂-GA with outstanding electrochemical performance as anodes for lithium ion batteries.⁹¹ 

Tadyszak et al. reported the synthesis and characterization of transition metal ion (TMi) doped partially reduced GO aerogels using VCl₃, CrCl₃, FeCl₂·4H₂O, CoCl₂, NiCl₂, and CuCl₂ chlorides as reducing agents (Figure 1.18).⁹²  The authors studied the influence of different TMis on the oxygen concentration and specific surface area of the derived aerogel, concluding that VCl₃ possesses the strongest reducing properties, resulting in the formation of the densest aerogel with the lowest oxygen content and lowest specific surface area.⁹² 

Chu et al. synthesized Ni, Co, and Mn doped SnS₂-GAs using metal chlorides and thioacetamide as precursors.⁹³  All the metal-doped SnS₂-GAs showed improved electrochemical performance compared to SnS₂. Mn-SnS₂-GA exhibited almost three times higher specific capacitance than SnS₂ (523.51 F g⁻¹ at the scan rate of 5 mV s⁻¹) and excellent cycling stability (98.57% capacitance retention after 2000 cycles at 10 A g⁻¹).⁹³ 

The first CNT aerogel was synthesized from aqueous-gel precursors in 2007 via critical-point drying and freeze-drying.⁹⁴  However, the mechanical integrity of pure CNT aerogels relied solely on van der Waals interactions,⁹⁴  which opened the possibility of investigating ways to increase their mechanical properties.

For example, Bryning et al. reinforced the CNT aerogel's network by adding 1 wt% of polyvinyl alcohol (PVA). Although a decrease in the conductivity of the CNT aerogel was observed, the authors achieved a significant increase in the mechanical properties (Figure 1.19).⁹⁴  Worsley et al., in an effort to increase the conductivity and the mechanical properties of CNT aerogels, changed the traditional polymer binder for a conductive binder by adding a CNT dispersion to a resorcinol-formaldehyde (RF) solution before gelation. SEM images of the carbonized CNT-carbon aerogel showed CNTs uniformly dispersed within the carbon aerogel matrix.⁹⁵  However, since the CNTs concentration was low compared to the concentration of RF, the improvements in the mechanical and electrical properties were modest.⁹⁴  A further reduction in RF concentration to 4 wt% and an increase in CNT to 2 wt% resulted in a CNT aerogel that simultaneously exhibited high electrical conductivity, mechanical stiffness, and super-compressibility.⁹⁶  Since then several reports have shown enhanced performance of CNT aerogels, though the recent focus has been on CNT-graphene hybrid aerogels.^29,97 

Sui et al. synthesized a CNT/graphene hybrid aerogel (CNT/GA) through heat treatment of aqueous suspensions of GO and CNT with dissolved vitamin C as a reducing agent, followed by supercritical CO₂ drying.⁹⁸  The CNT/GAs were investigated for the desalination of brackish water as capacitive deionization (CDI) electrodes, showing high removal capability for dyes and heavy metal ions including Pb²⁺ and Ag⁺. The hybrid aerogels exhibited a high conductivity of 7.5 S m⁻¹, a large BET (Brunauer–Emmett–Teller) surface area of 435 m² g⁻¹ with a hierarchically porous structure, and a desalination capacity of 633.3 mg g⁻¹ for a 35 g L⁻¹ NaCl solution.⁹⁸ 

Wang et al. synthesized a GA by hydrothermal treatment of GO in the presence of dopamine and FeCl₃. The GA served as a template for the in situ growth of CNTs to obtain a hybrid CNT/GA with enhanced surface area and hierarchical meso- and micro-scale pores.⁹⁹  This synthetic approach resulted in CNTs distributed within the layers of the GA (Figure 1.20), increasing the hydrophobicity, thermal stability, and oleophilicity towards organic compounds. The low-density CNT/graphene aerogel exhibited selective adsorption of organics and oils from water.⁹⁹ 

Fullerenes are characterized by their high electron affinity.^100,101  C₆₀ for instance is capable of storing up to 6 electrons in its triply degenerated lowest unoccupied molecular orbital (LUMO), which corresponds to 0.1 electrons per carbon.^100,102,103  For comparison, graphene can store ∼0.01–0.02 electrons per carbon within the electrochemical stability window of water.^104–107  Cerón et al. integrated fullerenes in graphene aerogels to improve the electrochemical activity of fullerenes by taking advantage of the high electrical conductivity and surface area of GAs.¹⁰⁸  The gravimetric current density of GA electrodes was increased upon physisorption of C₆₀ and C₆₀ monoadduct, which provided additional acceptor states in the form of the low lying LUMOs of C₆₀ and its derivatives. The hybrid GA-C₆₀ electrode showed ∼50% higher gravimetric peak current density than the pristine GA. Functionalization of GA with C₆₀-monoadduct doubled the gravimetric peak current density of the GA electrode (Figure 1.21).¹⁰⁸  Further optimization of this hybrid system can be achieved by covalently bonding fullerene derivatives to the graphene backbone, thus providing higher electrochemical stability.

GAs have been used in several 3D printing methods; here we summarize some of the most recent examples.

Direct ink writing (DIW), also known as robocasting, is an extrusion-based technique that involves the extrusion of ink through a fine nozzle, which is programmed to follow a toolpath that allows the construction of a 3D structure. The DIW technique employs a three-axis motion stage to assemble 3D structures by robotically extruding a continuous “ink” filament through a micronozzle at room temperature in a layer-by-layer scheme. The prerequisite for this method is to design gel-based viscoelastic ink materials possessing shear thinning behavior to facilitate extrusion flow under pressure and a rapid pseudoplastic-to-dilatant recovery resulting in shape retention after deposition. This technique was first adapted by Zhu and co-workers to form a 3D periodic microlattice of GAs.¹⁰⁹  Zhu et al. fabricated high concentrations of aqueous GO suspensions (20–40 mg mL⁻¹ GO) which exhibited shear thinning properties but lacked the stiffness to support its own weight while printing. To further enhance the stiffness and viscosity, hydrophilic silica particles were added to the ink. The 3D aerogels must remain wet during printing so that the liquid can be removed either by freeze-drying or supercritical drying to prevent the collapse of pores under ambient conditions. Therefore, the printing process is carried out in the presence of an organic solvent (isooctane) immiscible with the GO ink. After subsequent gelation at 80 °C followed by supercritical drying, the DIW aerogel is thermally reduced under an inert atmosphere at 1050 °C to recover the graphene properties. The leftover silica particles are removed by etching in the presence of hydrofluoric acid. The physical properties of the 3D printed GA are like those of the bulk GA (Figure 1.22).¹⁰⁹ 

The DIW aerogels had large surface areas (up to 1100 m² g⁻¹) and pore volumes (2–4 cm³ g⁻¹), and carbon: oxygen ratios above 20. The electrical conductivity of the DIW aerogels varied from 87 to 278 S m⁻¹. Further, the aerogel exhibited super-compressibility of up to 90% of the compressive strain. The Young's modulus vs. density of the bulk and printed GAs obeyed the power-scaling law (E ∝ ρ^2.5), indicating that the failure mechanism was mainly bending dominated for these cellular materials. Interestingly, the engineered microlattice displayed higher (almost twice) Young's modulus for a given density when compared to bulk GAs. The electrical resistance of the printed GA only slightly decreased under cyclic compression, confirming structural resilience. Highly stretchable aerogels were reported by Guo et al., by reinforcing GO inks with multi-walled carbon nanotubes. Aerogels with a 200% elongation through hierarchical synergistic assembly were printed using the DIW method.¹¹⁰  Zhu et al. also applied DIW GA for energy storage applications, such as supercapacitors. Through the addition of graphite nano-platelets (GNP) the electrical resistance of the DIW GA was lowered to ensure sufficient rate capability and capacitance of the DIW electrode.¹¹¹  The DIW GA electrode significantly outperformed its bulk GA counterpart and provides an example of how one can use 3D printed electrodes to overcome mass transport limitations and boost energy storage performance.¹¹² 

Yao and Chandrasekaran et al. further exploited the advantage of 3D printed graphene electrodes as a conductive scaffold/current collector to increase the mass loading of MnO₂, a pseudo-capacitive material. The capacitive performance of the 3D electrode is not limited by ion diffusion even at extremely high mass loadings, which is impossible for conventional bulk electrodes. Most importantly, these findings validate the concept of “printing” practically feasible pseudo-capacitor electrodes and devices.¹¹³ 

Jiang et al. also showed that printed GA can be structurally resilient and exhibit extraordinary capacitive rate and cycle performances.¹¹⁴  Responsive graphene inks were fabricated by García-Tuñon and Barg et al., using GO sheets chemically modified with a branched polymer as a precursor. Structures with low concentrations of GO could be printed with high resolution (100 µm) by converting the ink to a ‘pseudo-gel’ when the pH is <4. The modified GO sheets respond to lower pH by forming a noncovalent network with 5 orders of magnitude higher modulus than the ink at pH >8, making it printable via DIW.¹¹⁵ 

Zhang et al. achieved a low-density 3D printed GA (10 mg cm⁻³) by combining drop-on-demand inkjet printing with freeze-casting of GO suspensions.¹¹⁶  The technique involves rapidly freezing aqueous GO suspensions on a cold sink held at a temperature of −25 °C. Unlike the DIW method, low-viscosity Newtonian fluids can be used for drop-on-demand inkjet printing. As the printing progresses, every new layer is deposited onto an already frozen layer, which upon contact melts and the low-viscosity ink fills the voids between the layers and is re-frozen again as the whole structure is still in contact with the cold sink. This ensures good adhesion between the layers because of interlayer diffusion. The authors observed that the 2D GO sheets are well aligned along the freezing direction in the 3D printed GA. The 3D printed structure exhibited good structural integrity due to the bonding between the layers when characterized under in situ compression up to 50% strain (Figure 1.23).¹¹⁶ 

The printed GAs possessed conductivities of 2–15 S m⁻¹ as the density increased from 0.5 to 10 mg cm⁻³ and were electrically resilient when compressed multiple times. The relationship between Young's modulus vs. density of the printed GA exhibited a lower scaling index of 1.4 unlike the conventional monoliths which gave a value of 2.5. The authors attributed this to the designed 3D macroscopic hollow structures. The electromechanical properties were studied by monitoring the resistivity change as a function of compressive strain. 3D printed aerogels also exhibited remarkable super-elasticity over a temperature range from −100 °C to 300 °C.¹¹⁶ 

3D printing of pristine GAs using room temperature freeze gelation (RTFG) was first introduced by Lin et al.¹¹⁷  Like DIW,¹¹⁸  RTFG involves extrusion of ink through a micronozzle assisted by a programmed tool path to build a 3D structure.¹¹⁷  In RTFG, freeze gelation at room temperature is possible because the ink consists of graphene, not GO, suspended in an organic solvent, such as camphene or phenol, whose melting point is above 20 °C. High vapor pressure solvents were selected to facilitate sublimation at room temperature. The final architecture of the 3D printed GA mimics the traditional DIW, but the microstructure is determined by the solvent used. For instance, rGO inks prepared with phenol as a solvent resulted in aerogels with lamellar, directional based microstructures as observed for aqueous suspensions.¹¹⁹  On the other hand, camphene-based inks had rough interfaces more like metals, and since they solidify at room temperature, give the GAs a more random pore morphology (Figure 1.24).¹¹⁷ 

The printed GAs do not contain any chemical cross-linkers to boost their mechanical properties and hence are reinforced with polymers. However, the use of pristine graphene yields aerogels with large surface areas (up to 700 m² g⁻¹), high electrical conductivities (up to 9 S cm⁻¹) and densities of 20 mg cm⁻³. The RTFG aerogels also showed promising performance as electrochemical double-layer capacitor electrodes with energy densities as high as 27 W h kg⁻¹ and power densities up to 21 kW kg⁻¹, which is among the highest reported for 3D printed aerogels. Freeze-casting of GO was adapted by Wang et al.,¹²⁰  to form radial and centrosymmetric structure within GAs. Through controlled formation of ice crystals in aqueous GO dispersions, aerogels with aligned channels and predetermined pore sizes were obtained.

Complex shaped 3D graphene lattices were fabricated via the casting of GO suspensions in a 3D printed polymer mold by Zhang et al.¹²¹  In this study, GO/ethylenediamine (EDA) ink is poured into a hollow polymer architecture. The polymer mold with the desired wall thickness is generated through projection micro-stereolithography. To obtain the 3D GAs, the GO/EDA ink is hydrothermally reduced, freeze-dried and the polymer template removed via thermal etching (Figure 1.25). These hydrophobic GA lattices had densities as low as 1.6 mg cm⁻³, BET surface areas of ca. 15 m² g⁻¹ and electrical conductivities of 11–81 S m⁻¹.

When compressed, the unit cells of GAs exhibited elastic bending and compression deformation until reaching an elastic strain of 4%. Beyond that structure showed local yielding and fracture propagation until the whole structure collapsed. This work also investigated the chemical sensing capabilities of the 3D GAs and stated that these aerogels have a sensitive detecting response to acetone. When tested for other organic solvents, the detection sensitivity varied from 5% to 22%. In addition, the GA lattices showed potential sensitive chemiresistor properties for organic solvents, absorption capacity toward solid organics, such as asphalt, and good cycling stability.

As noted above, many techniques have been used to print 3D aerogels with moderate structural control; however, they have all failed to directly demonstrate a truly arbitrary design space primarily due to the limits of the printing technique (e.g., toolpath requirement, casting, and serial writing). Thus, many of the 3D printed aerogels are still limited in their design and minimum feature size (>100 µm). To address this issue, Hensleigh et al. reported a process to 3D print GAs with essentially any desired architecture and resolutions an order-of-magnitude finer than any previously reported using a light-based 3D printing technique called projection micro-stereolithography (PµSL).¹²²  The key breakthrough of this technique was the development of photocurable GO resins that (i) rapidly solidify by light-initiated polymerization, (ii) have strong light absorption to maintain small (µm-scale) layer thicknesses, and (iii) have sufficiently low viscosity to allow dipping and recoating for the layer-by-layer processing (Figure 1.26).¹²² 

The resin is a dilute (1 wt%) GO dispersion with a small amount of photocurable acrylates (12 wt%) and photo-initiator (2 to 4 wt%). The GO suspension consisted of crosslinked GO particles (XGO) made by ultrasonically dispersing a GO hydrogel monolith. It was shown that the crosslinked GO in the XGO resin led to higher surface area aerogels than simply using neat GO flakes in the GO suspension. The acrylates and initiator are the photoactive elements that allow PµSL printing by forming a temporary “green” structure that traps the XGO in the desired 3D architecture. The majority of the resin is solvent, N,N-dimethylformamide (DMF) as it provides a stable GO suspension, and solubilizes the acrylates and photo-initiator. The green structures are kept in solvent until dried either by supercritical or freeze-drying processes to maintain surface area. Pyrolysis of the “green” structures eliminates the majority of the photopolymer and reduces the GO, yielding the complex hierarchical 3D micro-architected graphene (MAG) assemblies (Figure 1.27).¹²² 

Using this technique, 3D printed GAs could be extended to gyroid and octet-truss geometries with features as small as 10 µm. Specifically, MAG octet-truss geometries exhibited much improved elastic moduli at low densities in addition to high surface areas and good electrical conductivities. MAG's essentially unlimited design space, high surface area and electrical conductivity opens the path to exploring mesoscale architectures for advanced GA applications including catalysis and separation platforms, tunable thermal conductivity, and fluid flow.¹²²  Similarly, Feng and Li et al. successfully fabricated graphene reinforced poly nanocomposite via digital light processing (DLP), showing the great potential of current photocurable resins. 3D complex structures including a jawbone with a square architecture as well as gyroid scaffold for bone tissue engineering applications were printed.¹²³ 

FDM is the most common 3D-printing technique used to create complex 3D objects layer-by-layer by extruding molten plastic filaments (or metal wires) through a nozzle, while the nozzle moves along the three axes via a computer-controlled mechanism such as DIW. This is typically a low precision technique which needs surface finishing treatment after extrusion and hardening, but offers other advantages, such as low fabrication costs and large-scale printing. The two most widely used filaments in FDM are acrylonitrile-butadiene-styrene (ABS) and polylactic acid (PLA). Wei et al. first reported 3D printing of a graphene/ABS and graphene/PLA composite using this technique.¹²⁴  Graphene-ABS composites were prepared with different graphene loadings and were further extruded into 17.5 mm diameter filaments (Figure 1.28). These filaments were extruded through a 0.4 mm diameter nozzle onto a platform held at 80 °C. When the graphene loading exceeded 5.6 wt%, inhomogeneity and aggregation of graphene flakes resulting in clogging of the nozzle being observed. The highest graphene-loaded printable composite (5.6 wt%) possessed an electrical conductivity of 1.05 × 10⁻³ S m⁻¹.

Foster et al. 3D printed free-standing electrodes using graphene-based mPLA filaments in a conventional FDM printer.¹²⁵  The authors also built a 3D printed solid-state supercapacitor (3D-SC) to evaluate the potential of this 3D printable graphene filament for energy storage. Utilizing two 3D printed discs with only 8% graphene loading and a solid electrolyte sandwiched between the discs, a fully free-standing supercapacitor could be created. A columbic efficiency of 85% was observed after 120 cycles, with an irreversible capacity reaching 3.69 mA h g⁻¹ with respect to the weight of the 3D printed electrodes.¹²⁵ 

Another promising application of these 3D printed electrodes was the hydrogen evolution reaction (HER) where a low overpotential for HER onset was observed even after the 1000th scan, thus making it the most beneficial electrode towards the HER of all the carbon-based electrodes examined.¹²⁵  Pumera et al. also reported a simple activation method for graphene/polymer 3D printed electrodes by a combined solvent and electrochemical route on a graphene/PLA filament.¹²⁶ 

There are a number of laser-based techniques that have been used to 3D print carbon structures. Laser induced graphene (LIG) is formed by irradiating a carbon source with a laser, which photothermally converts the carbon to sp²-hybridized carbon.¹²⁷  3D LIG foams are fabricated by preparing layers of LIG through the irradiation of polyimide (PI) film. Initially, the PI film is exposed to a CO₂ laser to form the first layer of LIG. The LIG layer is then coated with ethylene glycol (EG) which acts as a binding agent, before the next PI film layer is deposited. The sandwiched layers are then lased, and the process is repeated to build macroscale foams of LIG. After the process is complete to the desired height, the printed foam is dried at 200 °C to evaporate the remaining EG (Figure 1.29). Alternatively, the foam can be dried in a vacuum at 600 °C to remove any excess polymer residue.¹²⁷ 

Sha et al. used selective laser melting (SLM) to fabricate a Ni/sucrose scaffold using a CO₂ laser, where sucrose serves as a carbon precursor and Ni acts as a catalyst and template to form a free-standing 3D GF.¹²⁸  In situ synthesis of free-standing 3D GFs was successfully modeled by manually placing a mixture of Ni and sucrose onto a platform and then using a commercial CO₂ laser to convert the Ni/sucrose mixture into 3D GFs, the Ni metal catalyzing the graphene growth from the sucrose carbon source. This technique combines powder metallurgy templating with 3D printing techniques (Figure 1.30).¹²⁸  The 3D printed GFs show high porosity (∼99.3%) and low density (∼0.015 g cm⁻³). The GFs have an electrical conductivity of ∼8.7 S cm⁻¹, a remarkable storage modulus of ∼11 kPa, and a high damping capacity of ∼0.06. The printed 3D GFs showed minimal shrinkage of 20% in width. Shrinkage can be tuned by varying the size of Ni precursor and by selecting different carbon precursors. The two critical parameters in this technique are the laser duty cycle and the rastering speed. High quality graphene was obtained by using a fixed raster speed and high duty cycle.

Apart from the well-known techniques above, other methods such as stamping, templating, and laminated object manufacturing are also used to form graphene-based electrode materials in 3D patterns. For example, Zhang et al. demonstrated rapid production of flexible micro-super capacitors (MSC) through a scalable, low‐cost stamping strategy, wherein the authors combined 3D printed stamps and 2D titanium carbide or carbonitride inks (Ti₃C₂T_x and Ti₃CNT_x, also known as MXenes)¹²⁹  The viscous, aqueous MXene inks (24 mg mL⁻¹) are brushed onto 3D printed stamps of various electrode designs and printed on paper substrates to produce the different coplanar MSC electrodes. After deposition of the PVA/H₂SO₄ electrolyte on the electrodes, the MSCs exhibited good aerial capacitance. For scaling up the production of supercapacitors, the authors demonstrated a roll-to-roll compatible method that could create dozens of MSCs in seconds (Figure 1.31).¹²⁹ 

Binder jet printing has also been used by researchers to 3D print thick graphene electrodes (300 µm) for supercapacitors.¹³⁰  Thermally reduced GO powders densified by adding acetone are spread on the feed bed and the motion-controller was pre-adjusted to a layer height of 100 µm. The dimensional resolution of these structures is limited by the penetration of injected binder to adjacent powders to ∼1 mm. Supercapacitor electrodes decorated with palladium particles printed using this technique demonstrated high gravimetric and areal capacitance values of 260 F g⁻¹ and 700 mF cm⁻², respectively.¹³⁰ 

The development of advanced multifunctional materials for energy and environmental applications is becoming increasingly important from the perspective of sustainable development. Attributing to their structural integrity and interconnected porosity, graphene aerogels manifest extraordinary nanoscale effects that result in new material systems with superlative properties and novel functionalities. With advancement in processing techniques, rational tuning of the materials properties of such graphene-based 3D macrostructure is the key to enhancing their energy and environmental performance. The extraordinary properties of these new materials will further stimulate the next generation of technologies in the fields of energy storage, filtration and separation, catalysis and sensors, among others.

This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.