Analysis of Porous Structure Parameters of Biomass Chars Versus Bituminous Coal and Lignite Carbonized at High Pressure and Temperature — A Chemometric Study

The characteristics of the porous structure of carbonized materials affect their physical properties, such as density or strength, their sorption capacity, and their reactivity in thermochemical processing, determining both their applicability as fuels or sorbents and their efficiency in various processes. The porous structure of chars is shaped by the combined effects of physical and chemical properties of a carbonaceous material and the operating parameters applied in the carbonization process. In the study presented, the experimental dataset covering parameters of various fuels, ranging from biomass through lignite to bituminous coal, and chars produced at 1273 K and under the pressure of 1, 2, 3, and 4 MPa was analyzed with the application of the advanced method of data exploration. The principal component analysis showed that the sample of the highest coal rank was characterized by lower values of parameters reflecting the development of the porous structure of chars. A negative correlation was also observed between the carbon content in a fuel and the evolution of the porous structure of chars at high pressure. The highest total pore volume of chars produced under 1 and 3 MPa and the highest micropore surface area under 3 MPa were reported for a carbonized fuel sample of the highest moisture content.


Introduction
The operating parameters of the pyrolysis process, such as temperature, pressure, heating rate, and residence time, have been extensively studied in terms of their effects on the quality and yields of process products [1][2][3][4][5][6].These parameters have also been reported to influence the swelling behavior and fluidity of a fuel [3,[7][8][9][10] as well as reactivity of the resulting chars [4,9,[11][12][13].In some studies, the properties of the porous structure of chars, reflected in the total pore volume, pore area, and pore size distribution, were studied, since the shape and interconnections between pores significantly influence the physical properties of a material, such as its density, mechanical strength, and sorption capacity as well as thermal conductivity and mass flow in thermochemical processing [14,15].The influence of pyrolysis temperature and heating rate [5,[16][17][18][19][20][21][22], fuel particle size [17,22,23], as well as physical or chemical pre-treatment of carbonaceous materials [24][25][26][27] on the porous structure of chars has also been reported.In a limited number of studies, the effects of pressure on char development were considered [3,8,9,12,[28][29][30].These were reported to be complex, and also related to other parameters, i.e., coal rank, process temperature, heating rate, and residence time.The rise in pyrolysis pressure was reported to hinder [9,31] or enhance [5,12] the development of the surface area of chars.A trend of increase in the specific surface area of chars with pressure, followed by a decrease with further pressure rise was also observed [5,28,29].Similarly, the influence of temperature on the development of porous structure of chars presented in the literature is not unambiguous.In general, higher pyrolysis temperature was observed to result in a higher value of the specific surface area and total pore volume [21,22].Some studies, however, showed that the development of the micropore volume or area, the specific surface area, and the total pore volume may deteriorate when a certain pyrolysis temperature value is exceeded [13,21,26,30].These relations may be even less certain when a joint effect of elevated temperature and pressure is to be considered.In the study presented, the combined effects of high pressure and temperature on the evolution of char porosity was analyzed on the basis of the experimental results, and with the application of a chemometric method of data exploration.Various carbonaceous materials, ranging from bituminous coal through lignite to biomass, were tested in order to take into account the influence of a parent material composition on the porous structure of carbonized materials.The study presented is considered to narrow the research gap defined above, and to contribute to the studies on properties of carbonized materials produced under pressure in slow pyrolysis conditions, applicable in thermochemical processing and sorption processes.
Chars were produced with the application of a high-pressure thermogravimetric analyzer with the magnetic suspension balance mechanism (Rubotherm GmbH, Bochum, Germany) (see Figure 1).The analyzer enables thermogravimetric measurements on samples that are approximately two orders of magnitude larger than conventional thermogravimetric analyzers under atmospheric or pressurized conditions.It is equipped with an automatic control and indicating system, including automated gas dosing, data acquisition, and recording.A fuel sample of 1 g was pressurized in the argon atmosphere to 1, 2, 3, or 4 MPa, heated with the heating rate of 20 K/min to 1273 K, and processed under the final pressure and temperature conditions for 5 h.Relatively low heating rate and long residence time were applied, representing in situ coal processing conditions, and slow pyrolysis conditions applicable in the production of carbonized materials for thermochemical conversion or sorption purposes.
Energies 2017, 10, 1457 3 of 10 pressurized conditions.It is equipped with an automatic control and indicating system, including automated gas dosing, data acquisition, and recording.A fuel sample of 1 g was pressurized in the argon atmosphere to 1, 2, 3, or 4 MPa, heated with the heating rate of 20 K/min to 1273 K, and processed under the final pressure and temperature conditions for 5 h.Relatively low heating rate and long residence time were applied, representing in situ coal processing conditions, and slow pyrolysis conditions applicable in the production of carbonized materials for thermochemical conversion or sorption purposes.

Porous Structure Characterization
Chars were crushed and sieved to the fraction below 200 μm and degassed under vacuum at 393 K overnight [15].The porous structure characterization in a wide range of meso-and micropores was performed with the application of the Autosorb iQ analyzer (Quantachrome Instruments, Boynton Beach, FL, USA) equipped with a high vacuum system with a turbomolecular pump and a low-pressure transducer [32].A wide range of meso-and micropores (diameter of 0.35-40 nm) was characterized qualitatively and quantitatively in a single measurement.The nitrogen sorption isotherm data at 77 K was applied in the determination of the specific surface area based on the multipoint Brunauer-Emmett-Teller (BET) method [33], pore size distribution based on the Density Functional Theory (DFT) method [34], and micropore volume and surface area based on the V-t-deBoer method [35].The total pore volume was quantified as the volume adsorbed at the relative pressure of 0.99.

Exploratory Analysis of the Studied Multivariate Dataset
In the chemometric analysis of the experimental data, principal component analysis (PCA) was applied [36][37][38][39][40].It is one of the exploration methods most commonly applied in multivariate data analysis.PCA allows reducing the data dimensionality, and enables its visualization and interpretation.It decomposes the initial data organized in a matrix X(m × n) into two matrices, S(m × fn) and D(n × fn), called the score and loading matrices, respectively: where m and n denote the number of objects and parameters, respectively, fn denotes the number of significant factors called the principal components (PCs), and R is the residuals matrix.The score and loading matrices are orthogonal.The columns of the score matrix and rows of the loading matrix are called the principal components (PC), or eigenvectors.Each PC is constructed as a linear

Porous Structure Characterization
Chars were crushed and sieved to the fraction below 200 µm and degassed under vacuum at 393 K overnight [15].The porous structure characterization in a wide range of meso-and micropores was performed with the application of the Autosorb iQ analyzer (Quantachrome Instruments, Boynton Beach, FL, USA) equipped with a high vacuum system with a turbomolecular pump and a low-pressure transducer [32].A wide range of meso-and micropores (diameter of 0.35-40 nm) was characterized qualitatively and quantitatively in a single measurement.The nitrogen sorption isotherm data at 77 K was applied in the determination of the specific surface area based on the multipoint Brunauer-Emmett-Teller (BET) method [33], pore size distribution based on the Density Functional Theory (DFT) method [34], and micropore volume and surface area based on the V-t-deBoer method [35].The total pore volume was quantified as the volume adsorbed at the relative pressure of 0.99.

Exploratory Analysis of the Studied Multivariate Dataset
In the chemometric analysis of the experimental data, principal component analysis (PCA) was applied [36][37][38][39][40].It is one of the exploration methods most commonly applied in multivariate data analysis.PCA allows reducing the data dimensionality, and enables its visualization and interpretation.It decomposes the initial data organized in a matrix X(m × n) into two matrices, S(m × f n ) and D(n × f n ), called the score and loading matrices, respectively: where m and n denote the number of objects and parameters, respectively, f n denotes the number of significant factors called the principal components (PCs), and R is the residuals matrix.The score and loading matrices are orthogonal.The columns of the score matrix and rows of the loading matrix are called the principal components (PC), or eigenvectors.Each PC is constructed as a linear combination of original variables with weights maximizing the description of the variance of the data (i.e., S = XD').
Moreover, the sum of the squared elements of each PC is called an eigenvalue and represents the portion of the variance described by the corresponding PC.Because the first PC describes the largest amount of the data variance, the associated eigenvalue also has the highest value.The sum of the eigenvalues defines the total variance of the data.Providing that the reduction of data dimensionality is effective, it is possible to apply score vectors and loading vectors (i.e., the columns of a matrix S and rows of matrix D', respectively) to visualize and interpret the relationships between the objects and the parameters in a matrix X.

Experimental Dataset
The studied experimental dataset was organized into a matrix X(6 × 46).The rows of the matrix X represent studied fuel samples-objects 1-6, whereas the columns correspond to the studied parameters listed in Table 2.These cover physical and chemical parameters of fuels tested as well as parameters describing the porous structure of chars produced at high temperature and under various pressure conditions.The dataset was centered and standardized before the PCA model was constructed, since it included measurements performed within different magnitude ranges [39].

Results and Discussion
The effects of pressure applied in pyrolysis on the development of the porous structure of chars were observed for all samples.The values of the specific surface area, micropore volume and area, and the total pore volume are given in Figures 2 and 3, respectively.
The previously observed general trend of enhanced development of porous structure with the increase in pressure to the value specific for a given char was also reported, followed by a decrease in the specific surface area and total pore volume with a further increase in pressure [5,28,29].Under the atmospheric pressure conditions, the most developed porous structure was characteristic for lignite chars (see Figure 4).The value of the pyrolysis pressure corresponding to the most developed specific surface area and total pore volume of chars varied for various precursors and amounted to 2 or 3 MPa for lignites and bituminous coals, and 3 or 4 MPa for biomass.The highest values of the specific surface area and the total pore volume were reported for bituminous coal char (object 6) and lignite char (object 1) produced under 2 MPa, respectively, and amounted to 409 m 2 /g and 0.402 cm 3 /g.The highest increase in the specific surface area and the total pore as well as micropore area and volume Energies 2017, 10, 1457 5 of 10 of chars with pressure applied in pyrolysis were reported for bituminous coal (object 6) with a pyrolysis pressure of 2 MPa.The values of these parameters doubled when compared to the respective values under atmospheric pressure pyrolysis.
Energies 2017, 10, 1457 5 of 10 specific surface area and the total pore volume were reported for bituminous coal char (object 6) and lignite char (object 1) produced under 2 MPa, respectively, and amounted to 409 m 2 /g and 0.402 cm 3 /g.The highest increase in the specific surface area and the total pore as well as micropore area and volume of chars with pressure applied in pyrolysis were reported for bituminous coal (object 6) with a pyrolysis pressure of 2 MPa.The values of these parameters doubled when compared to the respective values under atmospheric pressure pyrolysis.The principal component analysis was applied in identification of the relationships between physical and chemical characteristics of the fuel samples, and the parameters describing porous structure of their chars under various pyrolysis conditions.The PCA model was constructed for the standardized studied data Xc(6 × 46).The data compression was effective and three principal components described 95.45% of the total data variance.The respective score and loading plots are presented in Figure 5.The principal component analysis was applied in identification of the relationships between physical and chemical characteristics of the fuel samples, and the parameters describing porous structure of their chars under various pyrolysis conditions.The PCA model was constructed for the standardized studied data Xc(6 × 46).The data compression was effective and three principal components described 95.45% of the total data variance.The respective score and loading plots are presented in Figure 5.The first two principal components described 83.73% of the total data variance.Along the PC1, presenting 60.63% of the total data variance, significant variations between objects analyzed could be noticed.The most significant differences were observed between lignite (object 1) and bituminous coal (object 5) and all the remaining fuel samples.Object 1 was characterized by a high content of moisture and ash in a sample (parameters 1 and 2), as well as high values of parameters describing the porous structure of its chars: the specific and micropore surface area, DFT cumulative surface area, total pore volume and micropore volume, DFT method cumulative pore volume and the average pore diameter under 0.1, 1, 2, 3, and 4 MPa (parameters 12-46), as well as a low content of carbon and a low value of heat of combustion and calorific value (parameters 6, 10, and 11).The bituminous coal sample (object 5) was unique mainly because of the highest fixed carbon, carbon, and nitrogen content in a sample (parameters 4, 6, and 8), the highest heat of combustion and calorific value (parameters 10 and 11), and lower values of the remaining parameters.Furthermore, for object 1, the highest total pore volume under 1 and 3 MPa (parameters 22 and 36), micropore surface area under 3 MPa (parameter 34), and DFT method cumulative pore volume under 3 MPa (parameter 38) were observed.
The PC2, which described 23.10% of the total data variance, showed the differences between coal samples (objects 1, 2, 5, and 6) and biomass samples (objects 3 and 4).These differences were observed mainly in terms of the physical and chemical parameters of fuels tested.Lignite and bituminous coal samples (objects 1, 2, 5, and 6) were characterized by relatively higher values of ash, fixed carbon, sulfur, and nitrogen content in a sample (parameters 2, 4, 5, and 8) and lower values of volatiles, hydrogen, and oxygen content in a sample (parameters 3, 7, and 9) than the biomass samples.The first two principal components described 83.73% of the total data variance.Along the PC1, presenting 60.63% of the total data variance, significant variations between objects analyzed could be noticed.The most significant differences were observed between lignite (object 1) and bituminous coal (object 5) and all the remaining fuel samples.Object 1 was characterized by a high content of moisture and ash in a sample (parameters 1 and 2), as well as high values of parameters describing the porous structure of its chars: the specific and micropore surface area, DFT cumulative surface area, total pore volume and micropore volume, DFT method cumulative pore volume and the average pore diameter under 0.1, 1, 2, 3, and 4 MPa (parameters 12-46), as well as a low content of carbon and a low value of heat of combustion and calorific value (parameters 6, 10, and 11).The bituminous coal sample (object 5) was unique mainly because of the highest fixed carbon, carbon, and nitrogen content in a sample (parameters 4, 6, and 8), the highest heat of combustion and calorific value (parameters 10 and 11), and lower values of the remaining parameters.Furthermore, for object 1, the highest total pore volume under 1 and 3 MPa (parameters 22 and 36), micropore surface area under 3 MPa (parameter 34), and DFT method cumulative pore volume under 3 MPa (parameter 38) were observed.
The PC2, which described 23.10% of the total data variance, showed the differences between coal samples (objects 1, 2, 5, and 6) and biomass samples (objects 3 and 4).These differences were observed mainly in terms of the physical and chemical parameters of fuels tested.Lignite and bituminous coal samples (objects 1, 2, 5, and 6) were characterized by relatively higher values of ash, fixed carbon, sulfur, and nitrogen content in a sample (parameters 2, 4, 5, and 8) and lower values of volatiles, hydrogen, and oxygen content in a sample (parameters 3, 7, and 9) than the biomass samples.
The PC3, describing 11.72% of the total data variance, also showed the uniqueness of object 6, which could be attributed to the highest micropore area and volume under 2 MPa (parameters 27 and 30).
The PC3, describing 11.72% of the total data variance, also showed the uniqueness of object 6, which could be attributed to the highest micropore area and volume under 2 MPa (parameters 27 and 30).The PC3, describing 11.72% of the total data variance, also showed the uniqueness of object 6, which could be attributed to the highest micropore area and volume under 2 MPa (parameters 27 and 30).The loading plots revealed a positive correlation between: parameters 6, 10, 11 (carbon content in a sample, heat of combustion, and calorific value); parameters 3 and 9 (volatiles and oxygen

Figure 1 .
Figure 1.Schematic diagram of the high-pressure thermogravimetric analyzer.

Figure 1 .
Figure 1.Schematic diagram of the high-pressure thermogravimetric analyzer.

Figure 4 .
Figure 4. (a) Pore surface area and (b) volume of chars produced under atmospheric pressure.

Figure 5 .
Figure 5. (a) Score plots and (b) loading plots of principal component analysis (PCA) for centered and standardized Xc dataset.

Figure 4 .
Figure 4. (a) Pore surface area and (b) volume of chars produced under atmospheric pressure.

Figure 4 .
Figure 4. (a) Pore surface area and (b) volume of chars produced under atmospheric pressure.

Figure 5 .
Figure 5. (a) Score plots and (b) loading plots of principal component analysis (PCA) for centered and standardized Xc dataset.

Figure 5 .
Figure 5. (a) Score plots and (b) loading plots of principal component analysis (PCA) for centered and standardized Xc dataset.

Table 1 .
Chemical and physical parameters of fuel samples.

Table 2 .
Chemical and physical parameters of fuel samples.