Abstract
The use of biocarbon derived from renewable resources to substitute for petroleum-based carbonaceous materials in composites and other applications often requires size reduction. Biocarbon obtained by the pyrolysis of miscanthus was subjected to ball milling from 2 to 24 h. Particle analysis was performed by combining scanning electron microscope imaging and image-based particle counting. The milled biocarbon had a highly heterogeneous shape and size distribution, making image-based analysis the most suitable method. The average particle size was reduced from above 3 µm after 2 h of milling to below 1 µm after 24 h of milling. The specific surface area doubled from 148 m²/g to approximately 300 m²/g after 2 h of milling, but it did not change with longer milling. Ball milling caused a gradual decrease of the thermal conductivity from 0.137 to 0.116 W·m-1·K-1. The ash content increased from 8 to 17% after 24 h of milling. Polypropylene composites filled with the biocarbon with and without ball milling showed lower density and comparable mechanical properties to a talc-filled composite except for lower impact strength. Using ball milled biocarbon led to a steady increase of the impact strength with longer milling time, reaching values on a par with that of the talc composite.
Download PDF
Full Article
Sustainable Carbonaceous Biofiller from Miscanthus: Size Reduction, Characterization, and Potential Bio-composites Applications
Tao Wang,a Arturo Rodriguez-Uribe,a Manjusri Misra,a,b and Amar K. Mohanty*,a,b
The use of biocarbon derived from renewable resources to substitute for petroleum-based carbonaceous materials in composites and other applications often requires size reduction. Biocarbon obtained by the pyrolysis of miscanthus was subjected to ball milling from 2 to 24 h. Particle analysis was performed by combining scanning electron microscope imaging and image-based particle counting. The milled biocarbon had a highly heterogeneous shape and size distribution, making image-based analysis the most suitable method. The average particle size was reduced from above 3 µm after 2 h of milling to below 1 µm after 24 h of milling. The specific surface area doubled from 148 m²/g to approximately 300 m²/g after 2 h of milling, but it did not change with longer milling. Ball milling caused a gradual decrease of the thermal conductivity from 0.137 to 0.116 W·m-1·K-1. The ash content increased from 8 to 17% after 24 h of milling. Polypropylene composites filled with the biocarbon with and without ball milling showed lower density and comparable mechanical properties to a talc-filled composite except for lower impact strength. Using ball milled biocarbon led to a steady increase of the impact strength with longer milling time, reaching values on a par with that of the talc composite.
Keywords: Carbon; Biocarbon; Nano-particles; Milling; Sizing; Composite; Additives
Contact information: a: Bioproducts Discovery and Development Centre, Department of Plant Agriculture, University of Guelph, Guelph, Ontario N1G 2W1, Canada; b: School of Engineering, University of Guelph, Guelph, Ontario N1G 2W1, Canada; *Corresponding author: mohanty@uoguelph.ca
INTRODUCTION
Particulate carbon is extensively used in many commercial products including ink, rubber products, automotive parts, coatings, pigments, sensors, catalysts, and energy storage and conversion devices. Most particulate carbon materials, including carbon black, are produced by the incomplete combustion of heavy oil products such as tar. Due to the unsustainable nature of these materials, there have been increased research efforts in using carbon derived from renewable resources in applications such as carbon sequestration, and even in areas traditionally dominated by petroleum-based carbon (Peterson 2012; Aller 2016; Behazin et al. 2016). Biochar is the solid fraction obtained after biomass pyrolysis, which produces bio-oil and syngas as the liquid and gas products (Arnold et al. 2018). Biochar is rich in carbon, highly porous, and is traditionally used as a natural and low-cost soil amendment material (Amin et al. 2016).
Unlike petroleum-based carbonaceous materials that are generated from a highly viscous liquid raw material, biochar derived from biomass can retain some of the tissue and cell structures of the original plant. Reducing biochar to particulate biocarbon is essential for promoting this material as a substitute for conventional carbon products. Techniques that have been explored for reducing the size of biocarbon include ball milling and vibratory pulverizing (Abdullah and Wu 2009; Peterson et al. 2012, 2016). It has been reported that pyrolyzed biomass has much better grindability than the original biomass (Abdullah and Wu 2009). In the same study, the biocarbon was milled to the maximum of 4 min and the particle size was still of a few hundred micrometers.
The particle size along with the size distribution of a particulate material can have strong correlations with its physical and chemical properties, thus having a direct impact on its application and performances. A large number of methods have been developed to study particle sizes and size distributions. These methods can be roughly divided into three groups: separation methods (e.g., the simple sieving method and various sedimentation methods), counting methods (e.g., image analysis and time-of-flight counter), and beam methods (e.g., laser/light diffraction and scattering, focused-beam reflectance, and ultrasonic attenuation spectroscopy) (Naito et al.1998; Li et al. 2005).
When the size of the particles of concern is in micrometers and nanometers, the most commonly used methods have been image-based analysis and laser diffraction and scattering. Laser diffraction and scattering is a very popular sizing technique used in research and industry labs due to its fast analysis speed and ease of use. Newer instruments can measure not only solid particles suspended in a liquid medium but also dry powdery material directly. However, based on the principles of the analysis, laser diffraction and scattering method can only provide a reliable measurement of the particle size distribution when the particles are perfect spheres.
Blott and Pye studied the sizes of Ballotini glass beads and sand particles a few hundred micrometers in size using a laser-based technique, and compared the results to those obtained by sieving (Blott and Pye 2006). The laser-based analysis and sieving method produced the same results for the spherical glass beads, but gave very different values for the sand particles with a non-uniform particle shape.
While sieving in the example above is a separation method and can only give size ranges, Califice et al. (2013) analyzed particles with different shapes using 2D and 3D images and compared the results to those of laser-based analysis. By using 3D images as the most truthful representation of the particle geometry and size, they found that laser diffraction results present large errors when being used to analyze not only non-spherical particles but also mixtures of particles with different shape populations. Furthermore, besides the shape factor, the parameters entered for the laser diffraction and scattering, in particular the refractive index, can affect the results significantly (Beekman et al. 2005).
Image analysis is a counting method, meaning that the particles are inspected one at a time, and the results are the number-based average and distributions. For fine particles, microscopy combined with image analysis can fully reflect particle size, shape, and geometry. The disadvantage is that a sufficient number of particles need to be analyzed to have a comprehensive survey of a sample. This number can be very large, especially in the case of a wide particle size distribution. Image analysis software is often required. As shown later in this study, the morphology of biocarbon is highly heterogeneous, making image-based analysis the most suitable method of choice for sizing the material.
In the present work, biocarbon was produced from miscanthus by pyrolysis and then subjected to size reduction by ball milling. Scanning electron microscopy (SEM) and image analysis were combined to investigate the morphology and sizes of the biocarbon obtained with different ball milling time ranging from 2 h to 24 h. The size and shape changes were then correlated with the changes in the specific surface area, total pore volume, ash content, and thermal conductivity. Finally, in order to evaluate the performance of the biocarbon as a filler, biocarbon-filled polypropylene composites were fabricated, characterized, and compared to their talc-filled counterparts.
EXPERIMENTAL
Materials
The raw material used was miscanthus harvested in southern Ontario, Canada. The 1 to 3 cm chopped miscanthus containing 10% moisture was pyrolyzed in an oxygen-deprived environment in a continuous pyrolyzer (Metamag Inc., London, ON, Canada) with a built-in screw conveyor at 600 to 650 °C for 10 to 20 min until completely charred. The biochar obtained was ground in a hammer mill to pass a 1/64 in (0.4 mm) screen. Ball milling of the ground biochar was performed by using a Patterson Industries D-Type Ball Mill (Toronto, ON, Canada). The mill contained a ceramic-lined 60-ft long and 72-ft diameter vessel that was filled with 6,000 lbs of ceramic balls with diameters ranging from ¾ to 1½ in, and was operated at a rotational speed of 300 rpm. The pyrolyzed miscanthus biocarbon was milled for 2, 4, 12, 18, and 24 h, and subsequently labeled BcBM2h, BcBM4h, BcBM12h, BcBM18h, and BcBM24h, respectively.
For the preparation of the composites, homopolymer polypropylene, Grade 1120H from Pinnacle Polymers (Garyville, LA, USA), and Fusabond® 353P coupling agent from DuPont Company (Wilmington, DE, USA) were used as the matrix. The talc used in the comparison study was JetFil® P500 talc obtained from Imerys (Oakville, ON, Canada).
Methods
Particle morphology and size analysis
The particles were observed using a Phenom ProX scanning electron microscope of Phenom-World B.V. (Eindhoven, The Netherlands). The sample powder was deposited on carbon tabs adhered to aluminum stubs and imaged without coating. A series of images were taken at different magnifications to capture particles of different sizes. The images were then analyzed using the ParticleMetric particle size analysis application (version 1.0.346.44124) of Phenom-World B.V. to gather the shape and size data of the particles in each image. Because the particles did not have defined geometry, the circle equivalent diameter was used to represent the average size of the particles. Image taking and particle counting were continued until the number average particle size stopped changing noticeably with new additions of images/particles, which indicated that the average size of the particles counted was approaching the true average of the particles in the whole sample. The number of particles measured for each sample was typically more than 3,000.
Specific surface area and pore volume
The surface area and pore volume analysis of the particles was conducted based on the Brunauer–Emmett–Teller (BET) theory, by collecting the adsorption–desorption isotherms of nitrogen at 77 K using an Autosorb-iQ Gas Sorption System by Quantachrome Instruments (Boynton Beach, FL, USA). Each sample was vacuum outgassed overnight before being tested.
Ash content
The ash content was measured by using two different methods. In the first method, the ash was the residue left after the sample was heated in a furnace to constant weight at 750 °C, according to ASTM D1762 – 84(2013) Standard Test Method for Chemical Analysis of Wood Charcoal.
In the second method, the heating and weight measurement was taken using a thermogravimetric analyzer (TGA Q500, TA Instruments, New Castle, DE, USA). Approximately 10 mg of a sample was loaded, and the temperature was ramped at 20 °C/min under air purge to 900 °C and held at that temperature for 20 min. The temperature was increased to 1000 °C in the end, and the residual weight was taken as the weight of the ash.
Thermal conductivity and diffusivity
A Hot Disk TPS 500 Thermal Constants Analyzer from ThermoTest Inc. (Fredericton, NB, Canada) was used for the measurement. The sample powder was oven-dried at 105 °C until it reached a constant weight and cooled down to room temperature in a desiccator before being used for testing. The powder was loaded into the cylindrical cavity of the sample holder with a Kapton sensor (Dia: 6.378 mm) inserted in the middle of the powder mass.
Three replicate measurements were taken for each sample, at 120 mW heating power and 60 Hz frequency.
Composites processing and characterization
All fillers were dried to constant moisture content before extrusion. The composites were produced by extrusion and injection molding using a 15-cc micro compounder from Xplore Instruments B.V. (Sittard, The Netherlands). The extruder has three independently controlled heating zones and twin screws with a length of 150 mm and aspect ratio of 18. Compounding was done with a screw speed of 100 rpm, processing temperature of 185 °C in all heating zones, and residence time of 2 min. The samples were molded at 30 °C with an average injection pressure of 10 bar and holding time of 24 sec.
All composites contained 27% filler, 3% coupling agent (Fusabond® P353), and 70% PP (1120H) by weight. Four biocarbon composites were prepared using the biocarbon before ball milling and the materials ball milled for 2, 4, and 24 h (BcBM2h, BcBM4h, and BcBM24h). A talc composite with the same matrix and composition was made using the JetFil® P500 talc. The tensile and flexural properties of the composites were tested using a universal testing machine (UTS-3382 by Instron, Norwood, MA, USA). Tensile tests were conducted using Type IV specimens according to ASTM D638 at a crosshead speed of 5 mm/min.
Flexural properties were tested according to ASTM D790 with a crosshead speed of 14 mm/min and span distance of 52 mm. The Standard Test Method for Unnotched Cantilever Beam Impact Resistance of Plastics ASTM-Designation: D4812 was used for the Charpy impact resistance measurements. The Standard Test Method for Determining the Izod Pendulum Impact Resistance of Plastics ASTM-Designation: D256 was used for the notched Izod tests. The density was measured by using a MDS-300 densimeter by Qualitest (Lauderdale, FL, USA).
The morphology of the composites and the neat polypropylene was studied by imaging the fracture surfaces using the Phenom ProX scanning electron microscope. The sample bars were abruptly fractured in liquid nitrogen and observed without coating. Images were taken at an accelerating voltage of 10 kV.
RESULTS AND DISCUSSION
Particle Size Analysis
The pyrolysis of the miscanthus resulted in black colored char with no considerable size changes from the raw material. The char was hammer milled to pass a 1/64 in (0.4 mm) screen and then ball milled for different lengths of time to yield powders.
As shown in Fig. 1(a), the hammer-milled sample contained large chunky pieces, some of which still clearly represented the cell wall structure of the original biomass. The other images in Fig. 1 showed that ball milling noticeably reduced the particle size.
A number of observations can be made of the ball-milled samples. The sample ball milled for 2 h showed much smaller particle size compared to the hammer-milled sample. However, it still contained a large number of plate-like structures, which appeared to be broken pieces of the original cell wall. The number of these plate-like pieces was much reduced in the 4 h ball-milled sample and difficult to find in the samples milled for longer periods of time. Particles measuring more than one µm could be found in all the samples. Even after extensive ball milling (e.g., 18 h and 24 h), the sizes of these large particles were still comparable to that of the similar particles found in samples milled for shorter periods of time.
However, the images also showed that the particle assemblies grew more “crowded”, largely due to the increased presence of the small “dots”. Higher magnification images revealed that these dots were sub-micron particles and they tended to aggregate to form a cluster or to decorate the surface of larger particles (Fig. 1 d’, e’, and f’). All ball-milled samples contained sub-micron particles and the number of sub-micron particles increased with ball milling time. Furthermore, although most of the particles had irregular shape, the particles milled for a longer period of time were less likely to have sharp corners, i.e., they became more spherical.
Because the particles had an irregular shape, image-based analysis was the only reliable method to study their sizes. The sample before ball milling contained very large particles, as shown in Fig. 1(a), making it unsuitable to be sized using an electron microscope. In fact, the ParticleMetric application used in this study has an analysis range of 100 nm to 0.1 mm. Therefore, the size analysis was only performed for the ball-milled samples. Figure 2 shows an example of one image being analyzed. The colored picture on the right demonstrates the “particles” that were recognized. Although not all particles in each image are being recorded, and some overlapping can occur to a small number of particles, the effect becomes negligible when a large number of particles (typically above 3,000 in this study) are measured. The circle equivalent diameter, which is the diameter of a circle that would have the same area as a particle on an image, was used to represent the size of the particle.
The average sizes in Fig. 3 confirmed the size reduction with milling time. The average particle size was reduced from 3.2 µm after 2 h of ball milling to 1.1 µm after 18 h of milling and 0.9 µm after 24 h of milling. It is worth pointing out that if one uses the images in Fig. 1 to manually measure the particle sizes, one would likely obtain a higher number than the average calculated here because the larger particles can be more conveniently measured. Furthermore, as discussed earlier, image-based analysis is a counting method, and the average calculated is a number average.
The small particles, although less visible, were in much greater numbers than the larger particles. If these samples were analyzed using the laser diffraction and scattering method, one would expect a higher average size because the laser method calculates the volume average and even a large number of very small particles do not contribute to the volume noticeably.
Fig. 1. SEM images of the biocarbon before and after ball milling; the samples labeled BcBM2h, BcBM4h, BcBM12h, BcBM18h, and BcBM24 were milled for 2, 4, 12, 18, and 24 h, respectively; the image in (a), the sample before ball milling, was taken at a much lower magnification than the other images
Fig. 2. SEM image of one ball-milled sample; the color picture on the right illustrates the particles recognized from the SEM image on the left; the color codes help to visually differentiate the particles and have no meaning in the measurement
The particle size distribution from the same analysis is presented in Fig. 4. The sample that was ball milled for 2 h (BcBM2h) showed the widest size distribution among all of the samples. As for the 4 h milled sample, although the highest peak shifted to a much smaller size number, the distribution was still broad.
In fact, it has been shown in Fig. 1(c) and (c’) that this sample still contained plate-like pieces. Further increase of ball milling time to 12 h and 18 h clearly reduced the percentages of the large particles (Fig. 4). The 24 h milled sample showed the narrowest particle size distribution.
Minimal studies have been published on the ball milling of biocarbon. Abdullah and Wu (2009) used ball milling to grind pyrolyzed wood for 1, 2, and 4 min to enhance its properties for fuel applications. Electron microscope images showed irregular particle shape and image-based particle size analysis found that approximately 28% of the particles obtained by milling the 300 °C pyrolyzed sample for 2 min were less than 75 μm. The milling time was relatively short, and the particles obtained were still very big. In a different study, birchwood biocarbon was ground by vibratory pulverizing using a gyro puck mill (Peterson et al. 2016). Laser-based size analysis showed three distinctive peaks in the ranges of 100 nm to 200 nm, 0.2 µm to 1 µm, and 4 µm to 7 µm. This result corresponds to the observation made in the present work that a small number of big particles was still present even after prolonged milling. Laser-based size analysis has limitations in accurately sizing these types of materials.
In the present study, the pyrolyzed biomass retained some of the structures of the plant tissue and cell wall, which were broken down to smaller pieces during the initial stage of ball milling.
With continued milling, more and more of the bigger particles were broken apart to smaller ones, leading to a quick increase of the particle number and in turn a decrease of the number-average particle size.
Fig. 3. Average and median particle sizes of the ball-milled biocarbon obtained by image-based size analysis
Fig. 4. Particle size distribution of the ball-milled biocarbon obtained by image-based particle size analysis (note: Although many of the samples contained particles larger than 10 µm, the percentages of those particles are very small. The portions of the curves beyond 10 µm are not shown to make the peaks on the left more visible in the figure.)
BET Surface Area
The size reduction of a material can lead to changes in its specific surface area. Figure 5 shows that the specific surface area of the sample before ball milling was 148 m²/g and the values of the ball-milled samples were between approximately 270 m²/g and 315 m²/g. Longer milling time with further size reduction did not lead to greater surface area. The total pore volume had the same trend, having an initial increase and then levelling off, as shown in this same Fig. 5.
Charred biomass is known to be a porous material (Ahmad et al.2014; Aller 2016). Figure 1 has shown that the initial ball milling caused considerable structural changes to the material, making more surfaces accessible. However, the fractionation of the particles with prolonged milling may have happened primarily along the large pore channels, thus causing the total surface area to remain relatively constant.
Peterson et al. (2012) used ball milling to increase the BET surface area of biocarbon produced from corn stover. While the unmilled material had a surface area of 3 m2/g, the samples that were wet milled with solvents, including acetone and ethanol, for 6 h showed much higher specific surface areas, with the highest value reaching 194 m2/g. Same as in the present work, longer milling time was found to be not helpful in further increasing the surface area. Compared to that study, the higher surface area values achieved by ball milling in the present work may have benefited from the higher surface area of the starting material (approximately 150 m2/g). No size analysis was reported in the cited work.