Abstract
The microscopic pore structure of wood is an important factor that affects its macroscopic properties. In this study, an oak sample was used for pore structure characterization. X-ray computed tomography imaging was carried out, and the scanning results (a multi-layer two-dimensional planar image) were rendered using Avizo software (a three-dimensional stereo image). A digital image processing method was used to identify the characteristics of the three-dimensional pore structure features and to calculate the characteristic parameters, i.e., the porosity (volume/surface), pore area and volume, pore size distribution, and the connectivity. The 27 mm3 oak sample had the following characteristics: a pore size which ranged from 8.56 µm to 1262.84 µm; a pore volume of 1.01 × 1010 µm3; a pore area and volume porosity of 1.12 × 109 µm2 and 37.6%, respectively; a surface porosity range of approximately 36.1% to 39.1%; a pore diameter for axial connection ranging from approximately 164.57 µm to 1262.84 µm; and had corresponding proportions of the pore area and volume of approximately 74.4% and 67.3%, respectively. This information provided useful structural data for the construction of future models.
Download PDF
Full Article
Pore Structure Characterization of Oak via X-ray Computed Tomography
Liangyan Guo, Hongchao Cheng, Junfeng Chen, Wentao Chen, and Jingyao Zhao *
The microscopic pore structure of wood is an important factor that affects its macroscopic properties. In this study, an oak sample was used for pore structure characterization. X-ray computed tomography imaging was carried out, and the scanning results (a multi-layer two-dimensional planar image) were rendered using Avizo software (a three-dimensional stereo image). A digital image processing method was used to identify the characteristics of the three-dimensional pore structure features and to calculate the characteristic parameters, i.e., the porosity (volume/surface), pore area and volume, pore size distribution, and the connectivity. The 27 mm3 oak sample had the following characteristics: a pore size which ranged from 8.56 µm to 1262.84 µm; a pore volume of 1.01 × 1010 µm3; a pore area and volume porosity of 1.12 × 109 µm2 and 37.6%, respectively; a surface porosity range of approximately 36.1% to 39.1%; a pore diameter for axial connection ranging from approximately 164.57 µm to 1262.84 µm; and had corresponding proportions of the pore area and volume of approximately 74.4% and 67.3%, respectively. This information provided useful structural data for the construction of future models.
Keywords: Pore structure; Porosity; Connectivity; XCT; Digital image processing
Contact information: Key Laboratory of Bio-based Material Science & Technology (Northeast Forestry University), Ministry of Education, Harbin 150040 China;
* Corresponding author: zjy_20180328@nefu.edu.cn
INTRODUCTION
The primary reason for the large variation in the characteristics of wood can be attributed to its complex microscopic pore structure. Furthermore, the microscopic pore structure of wood is also an important factor that affects its macroscopic moisture migration, heat transfer, and mechanical properties. Therefore, from a basic research and engineering applications perspective, a systematic in-depth investigation of the structure of the wood pores was of considerable interest.
Techniques for the characterization of the pore structure of wood included mercury intrusion porosimetry (MIP), gas adsorption (GA), scanning electron microscopy (SEM), and nuclear magnetic resonance (NMR) spectroscopy (Wang and Ximing 2014).
Among these techniques, MIP and GA are the most widely used for the characterization of wood pore structures (Kojiro et al. 2009; Plötze and Niemz 2010; Chang et al. 2011; Vitas et al. 2019; Jang and Kang 2019). These methods use the non-wetting properties of either Hg (mercury) or gas. Information on the pore structure is obtained indirectly by detecting the total amount of Hg or gas that enters the porous wood. However, both techniques have some weaknesses. Due to the pressure that occurs from the mercury intrusion, a part of the pore structure is inevitably altered during the intrusion process. This results in experimental error for the tested sample. In addition, mercury is toxic, and there are safety risks associated with this method. The method destroys the sample, since the investigated sample cannot be reused. The gas absorption operation is based on the detection of changes in pressure and temperature. Small variations can greatly impact the measurement results, and this method can only measure the area near the entrance of the sample; this inevitably leads to experimental errors. Scanning electron microscopy has obvious advantages in terms of the qualitative characterization of the 2-D (two-dimensional) pore structure, but the possibility of quantitative analysis is limited (Fuhr et al. 2012; Blanco et al. 2017). Nuclear magnetic resonance analysis is based on the Gibbs-Thomson equation, which can measure closed cells and is easy to operate (Wang and Chang 1986; Zhou et al. 2018; Bertolini et al. 2019). However, the equation used for calculations with this method is dependent on multiple variables, and errors can easily occur. In addition, the measurement range is small, and the experimental conditions are harsh.
X-ray computed tomography (XCT) is a non-contact, non-destructive 3-D (three-dimensional) microstructure characterization technique for materials. This technique can characterize samples with 3D structures at different scales. It has been successfully applied to investigate a variety of porous materials (Corigliano et al. 2017; Borges et al. 2019; Wan et al. 2019; Zhu et al. 2019). In the wood science field, researchers have used XCT to successfully observe the shrinkage behavior of wood (Taylor et al. 2013). X-ray computed tomography has also been applied to research related to wood archaeology (Stelzner and Million 2015; Maher 2020). Many sensitive wooden artifacts, e.g., wooden coffins, have been investigated via XCT in order to perform nondestructive testing. Information on the overall structure of the wood and the distribution of the medullary ray were obtained via XCT. In a study of timber rings, XCT successfully measured the internal structure of the wood and the annual ring density (Bondarenko et al. 2018). In a report on the penetration of adhesives and wood, researchers observed the penetration of the binder in the cell wall of wood via XCT. The researchers also observed the gradient of the binder penetration within the cell wall structure (Paris and Kamke 2015; Paris et al. 2015). However, there have been few reports on the usage of XCT for applications related to the investigation of the pore structure of wood, especially for pore connectivity applications.
In this work, XCT was used to determine the three-dimensional pore structure of wood. Combined with digital image processing methods, the characteristic parameters, i.e., the porosity, pore area, pore volume, pore size distribution, and connectivity were quantitatively calculated. The obtained wood pore data will be useful for pore structure reconstruction and the expansion of its applications.
EXPERIMENTAL
Materials
The oak wood samples were collected from the Greater Khingan Range forest area for investigation. The oak wood sample was cut into a cube with approximate dimensions of 3 mm x 3 mm x 3 mm. The three samples used for the XCT analysis was taken from the junction between the earlywood and the latewood in the heartwood. Each surface of the sample was sanded with 1000 grit sandpaper (a particle size of approximately 25.4 µm), and then the sample was oven-dried before testing. In order to eliminate the influence of water evaporation on pore structure in the detection process, the sample was oven-dried before testing. The drying temperature was 103 C, and the time was 24 h.
Methods
XCT setup
The XCT setup used a nano Voxel 3000 (Sanying Precision Instruments Co., Ltd., Tianjin, China). The maximum sweep voltage and current used was 100 kV and 200 µA, respectively. The flat panel detector size was 250 mm × 200 mm. The resolution was less than 5 µm, the number of physical pixel units in the detector was greater than or equal to 4800 × 3200, and the maximum resolution of a single two-dimensional reconstructed slice was greater than or equal to 12000 pixels × 12000 pixels. The distance between the sample and the ray source, and the ray source and the detector were 7.07 mm and 203.20 mm, respectively. The rotary scanning mode was used as the imaging mod, the exposure time for each image was approximately 1500 ms, and it took approximately 90 min to complete a scan; a Dell workstation (Precision 7530) was used as the computer. After scanning, the data were saved in a 16-bit-raw format with an original data size of 678 voxels × 777 voxels × 563 voxels (the size of a single voxel was 6.9 µm3).
Image processing
The image data obtained with the X-CT scan were digitized before performing image recognition and quantitative analysis. Generally, this process consisted of: (1) volume rendering; (2) image cropping; (3) noise filtering; (4) threshold segmentation; and (5) marking and quantitative analysis.
All of the above operations were performed with the 3-D visualization software Avizo (version 8.0, Thermo Fisher Scientific, Waltham, MA) and ImageJ (version 1.52a, National Institutes of Health, Bethesda, MD).
RESULTS AND DISCUSSIONS
Characterization and Analysis Reports
Figure 1 shows the X-CT scan of the three-dimensional pore structure and the two-dimensional section (cross section, radial section, and tangential section) of an oak sample. In Fig. 1, the black represents the pores and the gray represents the matrix. This image clearly reveals the main pore structures, such as vessels, wood fibers, and wood rays.
After a preliminary quantitative analysis of the above structures, the basic parameters of the corresponding pore structures could be obtained, as shown in Table 1. It was worth noting that the X-CT measurements were able to be used to determine the basic parameters, such as the volume porosity, pore volume, and area, as well as to obtain information that could not be detected via conventional methods, e.g., the surface porosity.
Table 1. Basic Parameters of Pore Structure from Oak
Note: porosity = Vpore / Vsample
As shown in Table 1, the pore size range of the oak samples was approximately 8.56 µm to 1260 µm, and the volume porosity was 37.6%. The porosity obtained in this study was lower than the 44.3%, which was determined via the mercury intrusion test in the study.
Fig. 1. The 3-D pore structure and 2-D slice image of oak wood. (a) and (b) are the 3-D pore structures, while (c), (d), and (e) are the cross, radial and tangential sections, respectively.
The sample used for this study was the same and the setup was an Auto-pore IV 9500 (Micromertitics, Atlanta, GA). This large discrepancy could have occurred due to the following three aspects. First, the resolution of XCT imaging was insufficient and only the micron scale porosity could be detected in this study; the nanometer level pores of the wood were not observable via XCT. Hence, the detected pore volume was less than the value measured via the mercury injection method. Second, the mercury intrusion method assumes connected cylindrical pores, so the experimentally determined porosity value is often higher than the actual value. Third, during the process of mercury intrusion, the microscopic cell wall of the wood was damaged due to the pressure of the mercury intrusion, which led to an increased amount of injected mercury. Therefore, the detection results would be artificially larger.
In addition to the volume porosity, the surface porosity of the different layers is shown in Fig. 2. Figure 2a shows the porosity distribution of each of the 513 layers, with a mean value of 37.6%, a standard deviation of 0.57%, and a fluctuation range of 36.0 to 39.1%. As shown by Fig. 2b, there was a certain difference in the surface porosity of each layer. This could be related to the natural variability of the wood samples. The difference in the surface porosity indicated that errors could occur when using two-dimensional SEM images to measure the surface porosity, and then using these images to approximate the volume porosity.
Fig. 2. Surface porosity from different slice layers. (a) slice along the z axis; (b) porosity from different slice
Quantification and Analysis of the Pore Structure Parameters
Figure 3 shows the three-dimensional pore structure that was obtained from different processing states after undergoing digital image processing, as follows: (a) was the original image; (b) was the pore and matrix image after filtering and threshold segmentation; (c) and (d) were the pore and the matrix images after separation, respectively; (e) was the pore structure of the axial connection; and (f) was the pore connectivity after screening. The pore aperture range of the axial connectivity was approximately 165 µm to 1263 µm, and the pore area and volume ratio were approximately 74.4% and 67.3%, respectively. During the detection process, only the axial pores were found to have connectivity, while the tangential and radial pores did not display connectivity. This phenomenon may be related to the definition of connectivity.
Fig. 3. The 3-D pore structure, (a) the original image, (b) the matrix and pore, (c) the matrix distribution, (d) the pore distribution, (e) the connectivity, and (f) the sieved connectivity
Broadly speaking, connectivity refers to a topological property of a space or a set. Connectivity implies a continuous process; however, in the pore structure of wood, there are very few pores that span the entire wood sample from top to bottom, and most of them are partially connected (Fig. 3b). The connectivity test in this study may have been too rigorous and formed only in axial pores, with no pores being formed in the other directions. It was shown in Fig. 3e and Fig. 3f that the transverse part showed the presence of interconnecting pores and a single color represented the same pore. Therefore, the definition of connectivity could be improved in the future, and the concept of “local connectivity” was proposed to more accurately assess and quantify pore connectivity.