Journal of Porous Materials

, Volume 19, Issue 5, pp 529–536

Novel methods to fabricate macroporous 3D carbon scaffolds and ordered surface mesopores on carbon filaments


    • Department of ChemistryUniversity College London
  • Lifeng Chen
    • Department of ChemistryUniversity College London
  • Negar Amini
    • Department of ChemistryUniversity College London
  • Shoufeng Yang
    • School of Engineering ScienceUniversity of Southampton
    • Department of ChemistryUniversity College London
  • Zheng Xiao Guo
    • Department of ChemistryUniversity College London

DOI: 10.1007/s10934-011-9501-x

Cite this article as:
Lu, X., Chen, L., Amini, N. et al. J Porous Mater (2012) 19: 529. doi:10.1007/s10934-011-9501-x


New methods for fabrication of 3D macroporous carbon scaffolds and synthesis of mesopores on carbon surfaces are proposed. Ordered macroporous filamentary carbon structures were made by rapid prototyping using solvent-based extrusion freeforming which allows the scaffold to be designed on computer and downloaded directly to a building platform. The surface of extruded filaments was decorated with 20–25 nm open mesopores by coating with nano-silica as a hard template followed by pyrolysis and dissolution of the silica. This left an open mesoporous surface to serve as a host for catalysts or enzymes while retaining integrity in the core for electrical and mechanical performance. The combination of these two methods could be used to make different hierarchical, multi-functional carbon structures which could be applied in fuel cells as the catalyst carrier and biofuel cell electrode.


MesoporesCarbonSilica colloidFilamentary macroporous structure

1 Introduction

Intricate special multifunctional carbon structures with a hierarchy of different pore distributions, each level of which serves a different purpose, are desirable for catalyst supports in reactors involving convective flow [1], for fuel cell electrodes [2] and in the development of biofuel cells [3] where they combine the roles of catalyst carrier, electrode and current carrier. Hierarchical porous carbon structures are therefore a goal for several research groups [4]. The sequence of operations involves construction of carbon scaffolds with different shapes, creating mesopores with different sizes and immobilization of catalysts or enzymes. The first two steps are addressed here.

In order to improve the mass and heat transfer and to control liquid or gas flow directions, spatial arrangements involving macropores of complex design are required and the procedure described here allows such designs to be prototyped. Traditionally, macroporous scaffolds have been fabricated by hard-template [5] or foaming methods [6]. Both are complex and the latter presents the difficulty of controlling the pore size and pore size distribution. In this paper, we propose the extrusion freeforming method to make 3D carbon structures. It makes use of a rapid prototyping method which is also capable of being used for manufacturing. The structure is designed on computer and downloaded directly to the building platform where it can deliver filaments down to 80 μm diameter [7] with inter-filament pore sizes down to ~5 μm. Extruded filaments of a paste consisting of polymer, fine particles (e.g. ceramic, metal) and solvent are assembled to construct nominally 2D or 3D structures. The solvent is propan-2-ol, sometimes used as a hand wash in the hospital environment. Extrusion freeforming has been successfully used for fabrication of EBG crystals [8] and artificial bone scaffold [7]. Compared with similar technologies, such as fused deposition [9] and gel extrusion [10], paste extrusion is a relatively simple technique without heating, cooling or polymerization processes to contend with and can be easily controlled. This technology offers a new method to fabricate electrode and catalyst/biocatalyst carrier scaffolds.

In order to immobilize an enzyme or catalyst on the carrier surface, a particular mesopore size range is required. For example, biofuel cell enzymes require supports with pore size around 15–35 nm [11]. However, the mesopores in the structure can be divided into two categories: accessible and inaccessible. Chemical catalytic reactions tend to take place on catalysts within multiphase fluids so the effective enzyme and catalyst hosts are mesopores situated on the accessible surface of the skeleton. If such pores are embedded within the structure, they are useless and can cause deterioration of mechanical and electrical properties. So a structural carbon supporting a mesoporous carbon film is the ideal electrode and catalyst host.

Structures of ordered porous carbon with one mode of mesopore size (normally 2–50 nm) have been reported. They are generally synthesized by nanocasting template [12] or self-assembly [13] methods. The ordering and spatial distribution are strongly dependent on the hard/soft templates. At present, there are difficulties in creating 20–25 nm surface mesopores. Traditional but effective soft-template methods using PVB (poly(vinyl butyral)) [14], Pluronic F127 (PEO106–PPO70–PEO106) [15] and FDU (Fudan University) mesostructures [16] which are suitable for preparation of well-ordered mesoporous carbon films can only make pores less than 10 nm. Other traditional methods using hard-templates like SBA-15 [17], MCM-48 [18], CMK-1 [19] are not suitable for coating to create mesopores on the substrate surface. In this paper, we propose the ordered macroporous carbon structure fabricated by extrusion freeforming and a new method for creating mesopores on the surface of carbon filament. The combination of these two methods could develop more complex hierarchical porous carbon structures.

2 Experimental details

2.1 Fabrication of ordered filamentary macroporous carbon

Ordered macroporous carbon lattice structures were fabricated by solvent-based paste extrusion freeforming in which the extruded structure solidifies by transition of polymer from liquid to gel through solvent evaporation accompanied by an overall increase in solids content. Two different kinds of paste were tested. The first was based on formulations previously used for ceramics [8]. The thermoplastic binder, poly(vinyl butyral) (PVB) (grade BN18, Whacker Chemicals, UK), and plasticizer, poly(ethylene glycol) (PEG) (MWt = 600, VWR, UK) were used at 75% wt. PVB, 25% wt. Graphite (GS6E, GrafTech International Ltd, UK) was incorporated at 60 vol.% based on the dry paste. A second paste was prepared from PVB, Novolac phenolic resin (Vilosyn 455, Vil Resins Ltd, UK) with a curing agent of hexamethylenetetramine (HMTA) (Reagent Plus 99%, Sigma-Aldrich, UK) with a mass ratio of PVB to resin 3:2 and graphite powder at 30 vol.% based on graphite + polymer + resin. In both cases, the solvent was propan-2-ol (General Purpose Reagent, VWR International, UK). In preparation, first PVB + PEG or PVB and resin + HMTA (w:w = 100:13.5) were dissolved independently in the solvent and then mixed. The graphite powder was added to the organic solution and dispersed by ultrasonic probe (U200S, IKA Labortechnik, Germany). The pastes were prepared by stirring the suspension manually while evaporating excess solvent to develop a paste suitable for extrusion. The viscosity of the paste was measured on a Haake cone and plate Reostress 150 rheometer.

The extrusion equipment is shown in Fig. 1a. There are four axes: X, Y, Z and extrusion. A stainless steel syringe (internal diameter 9 mm) was mounted on the Z -axis and a glass substrate was placed on the X–Y table. The syringe was fitted with an extrusion die with a nominal 230 μm diameter (Model 1423, sapphire waterjet cutting nozzle, Quick-OHM, Germany). The graphite powder is shown in Fig. 1b.
Fig. 1

a Schematic diagram of rapid prototyping extrusion equipment. b Scanning electron micrograph of the graphite powder (scale bar 10 μm). c Scanning electron microscope image of 20–25 nm silica particles

2.2 Synthesis of surface mesopores on extruded carbon filaments

The novel method for the production of 20–25 nm mesopores on the carbon filament surface is based on a coating of colloidal silica (Fig. 1c) applied before pyrolysis. The as-extruded carbon paste lattices were dipped in the colloidal silica suspension (Ludox AS-40, 40 wt% silica, size: 22 nm, Sigma-Aldrich, UK) to deposit a thin surface film and then dried at 50–60 °C. in air. Because PVB and resin are not soluble in water, the filaments are not damaged by this step. This process was repeated up to four times but these thicker coatings tended to spall during drying. In order to prevent spalling, PEG (Mwt: 10,000, VWR, UK) was added to the colloidal silica. The filaments were fully dried in an oven and the temperature was increased to 140 °C for 6 h to cure the resin. For carbonization, the filaments were placed in a metal wall furnace and nitrogen (oxygen-free, BOC, UK) was used as the protection gas. The temperature was increased from room temperature to 400 °C at 2 °C/min with a dwell at 400 °C for 1 h, then from 400 to 800 °C at 5 °C/min with a dwell at 800° for 0.5 h. Finally silica was leached using 5 M NaOH solution (NaOH: 97%, Sigma-Aldrich, UK) for 7 days at ambient temperature.

3 Results and discussion

3.1 Ordered filamentary macroporous carbon structures

The solid freeforming stage depends on the flow properties of the graphite paste and the way viscosity changes as evaporation of solvent proceeds. Conventional extrusion of graphite pastes has been used to make electrodes. Baldrianova et al. [20] used a graphite paste based on highly viscous silicone oil mixed with graphite powder at a volume fraction of about 56% in disposable micropipette tips to make electrodes within the diameter range from 50 to 300 μm. Cervini et al. [21] used graphite powder mixed with polyurethane resin to extrude 3.0 mm graphite rods. But these types of paste do not satisfy the requirements for extrusion freeforming because the extruded filaments need to bend 90° to stack layer by layer and then weld at junction points to the preceding layer. Paste extrusion freeforming depends on an effective state change brought about by the polymer transition from liquid to gel through solvent evaporation accompanied by an overall increase in solids content which takes place after deformation and welding.

The state change from liquid (in the extruder barrel) to gel (soon after extrusion) and to solid after complete drying, influences the quality of scaffolds made by extrusion freeforming. Initially, a conventional paste formulation for solid freeforming was tested based on experience with ceramic powders. Subsequently the formulation was redesigned. Figure 2 shows how the viscosities of these pastes changed with overall ceramic volume fraction at a fixed shear rate of 30 s−1. The traditional paste formulation presents no clear solid volume fraction at which there is a critical liquid to solid transition (circle symbols in Fig. 2); instead viscosity increases gradually as solvent evaporates. This means the extruded filament does not solidify quickly and defects such as sagging can occur between supporting filaments in the preceding layer. Indeed this paste could only be extruded through a nozzle with diameter greater than 0.5 mm and graphite structures fabricated with this paste contained alignment errors and sagging.
Fig. 2

Viscosity of pastes 1 and 2 at a shear rate of 30 s−1 as a function of the overall graphite volume fraction (including solvent). The graphs show how viscosity changes as solvent content changes and hence give an indication of the effect of evaporation

For the reformulated paste containing graphite powder, PVB, phenolic resin and solvent, the viscosity increased rapidly as the solvent content decreased reaching a critical value that indicates an effective state change (square symbols in Fig. 2). This paste was tolerant of finer extrusion nozzles and produced higher quality carbon structures as shown in Fig. 3a. The design dimensions of the structure shown in Fig. 3b are 10.35 × 10.35 mm. The smaller channel size is 110 × 110 μm and the larger channel size is 270 × 270 μm. In addition to the superior processing capability of paste 2, it has the advantage that Novalac resin has a high carbon yield. A yield of 70% for pyrolysis to 800 °C was recorded, comparable to values reported by others [22]. It is therefore preferred for fabrication of such porous graphite structures.
Fig. 3

Ordered macroporous carbon structures a fabricated by paste 2, b design plan showing two levels of macroporosity

Scanning electron micrographs of the extruded carbon lattice are shown in Fig. 4. Figure 4a shows a filament crossing and welded to the filament underneath. Figure 4b shows the lay-up precision of the design configuration given in Fig. 3b.
Fig. 4

SEM images of extruded carbon structures a filament, b carbon lattice

The advantages of ordered macroporous carbon fabrication by paste extrusion are: (1) Complex macroporous arrangements can be designed exactly according to the requirements of the porous carbon application and then constructed by the nozzle trajectory which is set by computer so that fluid flow and flow direction can be engineered into the final product; (2) No moulds or templates are needed; (3) The fabrication method is relatively simple, rapid and requires no heating, cooling or cross-linking. Various ceramic lattices designed for different scaffolds have been demonstrated and illustrate the scope of this procedure [21].

3.2 Surface mesopores on carbon filaments

The filaments of paste 2 which were used for coating are shown in Fig. 5. Figure 5a shows the filament surface after extrusion and drying but before binder removal. The surface is rough as a result of agglomerates of graphite powder. A fracture face of a filament that has been dried in ambient air after extrusion is shown in Fig. 5b. The filament retained its shape well, remaining circular in section and did not deform after extrusion and drying. Figure 5c shows the uncoated original filament with high magnification and provides a reference with which to compare the surfaces subsequently decorated with nano silica beads. At comparable magnification to that used for studying the mesoporous surfaces, it consists of smooth nodules in the 1–5 μm region.
Fig. 5

Scanning electron microscope images of filament of paste 2 a surface, b cross section (scale bar 100 μm). c Surface at high magnification

The colloidal silica was received in aqueous suspension and was difficult to redisperse uniformly in resin-based organic solutions of the carbon precursor. Initially, a mixture of colloidal silica and phenolic resin in ethanol dispersed by ultrasonic probe was used for filament coating but the silica beads tended to agglomerate in the resin–ethanol system. After carbonization, these agglomerates, some of them up to 20 μm in diameter, were embedded in the carbon as shown in a fracture surface of a cast layer of the coating system in Fig. 6a creating a much higher pore size than that sought. The PVB and resin are fully or partially soluble in many organic solvents so the coating suspension should ideally be water-based to preserve the filament itself from solvent attack and prevent it intruding into the coating to encapsulate the template. The templating silica beads on the surface should compact tightly and contact each other in order to provide a path for the leaching solution to dissolve them progressively from the surface inwards. If not, they become embedded in the carbon during carbonization of precursors and afterwards cannot be removed. An example of this effect was encountered in attempting to use colloidal silica dispersed in sugar solution as the coating medium. Figure 6b shows the surface that resulted after carbonization to 800 °C and the silica is seen to be partially agglomerated into 50–100 nm clusters. The carbon from sugar covers the silica beads so that when the sample was soaked in NaOH solution for 5 days, the silica beads were difficult to remove.
Fig. 6

Carbonized samples a colloidal silica applied by dispersing in phenolic resin–ethanol solution; the silica has agglomerated, b colloidal silica applied by dispersing in sugar solution so that the template is protected from subsequent dissolution

However if the extruded carbon was coated with a water-based silica suspension without additives in the form of sugar or resin, then the hard templates could deliver open 20–25 nm mesopores on the surface. Figure 7a shows the nearly close-packed arrangement of nanospheres on the surface after coating from water suspension and Fig. 7b shows that after pyrolysis and leaching in NaOH, this procedure leaves an open mesopore structure: agglomeration was avoided and encapsulation of the template did not take place. It should be noted that the magnification in Fig. 7a is nearly 1.5 times that in Fig. 7b: the mesopores resulting from alkaline leaching are around 20 nm on the surface.
Fig. 7

Filament surface at high magnification after a coating with silica template and b after drying and pyrolysis

Two transport mechanisms account for the final surface topography as seen in Fig. 7b. The main process takes place in the early stages of heating as the resin softens but before curing is completed and consists of capillary flow into the interstices of the assembly of nanospheres. During the subsequent heating and pyrolysis stages, infiltration of the voids between silica beads can also occur through the formation of pyrocarbons starting by nucleation and growth on the template silica beads [18]. The SEM images use high magnification close to the limit for this instrument (JSM 6301, JOEL, Japan). The original silica beads deposited from colloidal silica and dried are shown in Fig. 1c in order to gauge their dimensions for subsequent microscopy.

The distinguishing features of this method are: (1) The precursors of carbon come from the internal filament which is covered by a silica bead layer. (2) The direction of carbon deposition proceeds from internal to external. It prevents the carbon covering the silica beads and is therefore favorable for silica leaching. (3) The mesopores formed on the filament are open to the outside so it is easier to immobilize enzymes and catalysts and is therefore suitable to chemical reactions. If the carbon deposition direction was from external to internal, it would be difficult to leach the silica beads because they become embedded and protected by carbon.

The coating does not remain coherent during subsequent processing. If the coating of silica beads was too thick, outer sections of coating separated from the filament (Fig. 8) because of shrinkage during drying but a layer of beads remained over most of the surface. In order to prevent such peeling and to decrease the degree of cracking, 0.5 wt% PEG (polyethylene glycol, Mwt: 10,000) was added to the colloidal silica to improve the adhesion between particles. This prevented spalling but did not prevent cracks forming in the film. During subsequent pyrolysis, the film forms islands as shown in Fig. 9 but remains adherent. This is caused by expansion of the fibre during pyrolysis. The diameter of original carbon fiber before coating was 300 μm as shown in Fig. 5b. After coating with silica beads, the fibre diameter was 330 μm but after pyrolysis, the diameter of the fibre is 360 μm. This has two disadvantages; residual porosity inside the fibre and formation of islands in the coating. These issues remain as challenges to formulation and processing conditions to restrict expansion of the filament during pyrolysis and this might be achieved by using a higher volume fraction of graphite powder while retaining the same viscosity by using a wider particle size distribution or by modifying the curing process to give a higher crosslink density.
Fig. 8

Dried coating of silica deposited from aqueous suspension
Fig. 9

The coating has a tendency to separate from the filament during pyrolysis

The next stage in the development of these structure for fuel cells, which is a focus of a continued study within the UK Biological Fuel Cells Consortium, involves the controlled modification of the carbon surface e.g. by nitric acid [24] or ammonia [25] followed by immobilization of the enzyme (such as laccase) on the functionalized surface. The adsorption can be assessed by FTIR to obtain information about the immobilized enzyme and for comparison with those of free enzyme [26]. The activities of immobilized laccase can be evaluated by using ABTS in distilled water as the reaction substrate and one activity unit is defined as the amount of enzyme required to catalyze 1 μmol of substrate per minute [27]. During the reaction process, the increase in the absorbance at 420 nm can be measured by a UV–vis spectrophotometer [27]. Finally, the laccase-immobilized graphite electrode biofuel cell can be studied by cyclic voltammetry using a potentiostat [28]. This work has been conducted for similar modified carbons intended for fuel cell applications [29].

4 Conclusions

The use of extrusion freeforming makes it possible to build 3D carbon structures with hierarchical pore arrangements. The macropore structure can be designed on the computer to fulfill convective flow conditions specified, for example, by computational fluid dynamics modeling. Intricate carbon structures with a hierarchy of pore size distributions each level of which serves a different purpose could have applications in fuel cells and biofuel cells as the catalyst carriers and enzyme-immobilized electrode. Biofuel cell enzymes require pore sizes 15–35 nm so a structural carbon with mesoporous surface carbon film is the ideal electrode and catalyst host. Open surface mesopores in the required range were produced using a hard template coating but some problems still need to be resolved. The mesoporous skin of the filament can separate into islands during drying and pyrolysis. The mechanical strength of the filament needs to be improved and these problems are being addressed.


The authors are grateful for the support of the Leverhulme Trust under grant number F/07476/V and the EPSRC under the SUPERGEN Initiative via. the UK Biological Fuel Cells Consortium (Grant no. EP/H019480/1).

Copyright information

© Springer Science+Business Media, LLC 2011