Article pubs.acs.org/EF
Constitution of Drop-Tube-Generated Coal Chars from Vitrinite- and Inertinite-Rich South African Coals Enette B. Louw,†,‡ Gareth D. Mitchell,† Juan Wang,§,∥ Randall E. Winans,§ and Jonathan P. Mathews*,† †
John and Willie Leone Department of Energy and Mineral Engineering, The Earth and Mineral Sciences (EMS) Energy Institute, The Pennsylvania State University, 126 Hosler Building, University Park, Pennsylvania 16802, United States § X-ray Science Division, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, Illinois 60439, United States ∥ The Peac Institute of Multiscale Sciences, Chengdu, Sichuan 610207, People’s Republic of China ABSTRACT: The structural transformations of two Permian-aged South African coals, one vitrinite-rich [91.8% dry mineral matter free (dmmf)] and one inertinite-rich (87.7% dmmf), and their resultant char morphologies were compared in this study. With these two maceral-rich coals, the opportunity presented itself to compare the degree of thermoplasticity during coal-to-char formations for the macerals without the need to use maceral separation techniques. The thermoplasticity is affected by its petrographic composition and, consequently, influences the combustion behavior. Pyrolysis chars were generated from wetscreened coal (200 × 400 mesh), under rapid-heating conditions (104−105 °C/s) in a drop-tube reactor, to closely resemble chars generated in pulverized combustion conditions. The chemical and physical structures of the chars were characterized through a range of different analytical techniques, including scanning electron microscopy, X-ray diffraction (XRD), small-angle X-ray scattering (SAXS), nitrogen adsorption, and optical microscopy, to quantify the factors contributing to reactivity differences. Results indicated that the inertinite-rich coal experienced limited fluidity during heat treatment, resulting in slower devolatilization, limited growth in crystallite height (11.8−12.6 Å), and only rounding of particle edges and producing >40% of mixed dense-type chars. The vitrinite char showed more significant structural transformations, producing mostly (80%) extensively swollen crassisphere, tenuisphere, and network-type chars, and XRD showed a large increase in crystallite height (4.3−11.7 Å). Nitrogen adsorption revealed that both chars were mostly mesoporous but that the inertinite-rich char had double the average pore size, which also resulted in a larger nitrogen surface area (3.9 m2/g compared to 2.7 m2/g). SAXS data showed that the vitrinite-rich char had 60% higher frequencies of pores in the micropore range. Helium porosimetry indicated that the inertinite-rich coal and resultant char had higher densities than the vitrinite coal and char, 1.6 and 2.0 g/cm3 compared to 1.3 and 1.9 g/cm3 (on a dry basis). To evaluate combustion reactivity, non-isothermal burnout profiles were obtained through thermogravimetrical analysis in air. The burnout profiles showed that the inertinite-rich char had a burnout temperature of 680 °C, slightly higher than that of the vitrinite-rich char, of 650 °C. This along with the peak shape and position in the burnout profiles indicates that the vitrinite-rich char has a higher reactivity. The higher reactivity is due to a combination of factors, likely including less organization, greater porosity and access to the reactive site, less ash blocking, and char morphology differences. The char samples were de-ashed, which resulted in an increase in combustion reactivity, because the ash acted as a barrier to the reactive surface area. The maximum reaction rate of the high-ash (36% ash yield) inertinite-rich char increased 80% after deashing, while the vitrinite-rich char, with an ash yield of 15%, had a 20% increase in reactivity after de-ashing.
■
INTRODUCTION Char combustion is often limited by gas-phase diffusion and the rate of chemical reactions, depending upon parameters such as the temperature, reactor size, and particle size distribution. The degree of thermoplasticity during coal to char transition is dependent upon the maceral composition, rank, and heating rate.1 These factors affect the transformations to chars producing various char structures, morphologies, and hence reactivity.2−4 Macerals are the individual organic components that reflect differences in original plant material, depositional environment, and degree of coalification. Vitrinite is mainly from deposition of woody plant material (e.g., bark, wood, leaves, and roots), while inertinite is formed from plant material that has been oxidized prior to burial (e.g., decomposition caused by fire or aerobic bacterial decomposition).5 The maceral composition of a coal will influence reactivity6 because each maceral group has differences in chemical properties, such as aromatic structural order, oxygen content, and cross-linking, © 2015 American Chemical Society
that influence the degree of plasticity during heat treatment for the bituminous coal rank range. The distribution of the macerals within a particular coal can vary, which results in different structural parameters and, as a consequence, different combustion behaviors. Combustion of vitrinite-rich coals has been well-studied but less is known about the combustion behavior of inertinite-rich coals. These coals are used domestically in South Africa for electricity generation in specialty coal-fired boilers but not exported as a result of their lower reactivity than vitrinite-rich coals.7 Jones et al.8 studied the influence of the rank and maceral composition on char, where the porous structure from an inertinite-rich was compared. Using reflective light microscopy, it was found that chars derived from vitrinite produce Received: August 26, 2015 Revised: November 20, 2015 Published: December 2, 2015 112
DOI: 10.1021/acs.energyfuels.5b01517 Energy Fuels 2016, 30, 112−120
Article
Energy & Fuels Table 1. DTR Char Properties coal
Highveld char
Waterberg char
H
H1
H2
H3
H4
W
W1
W2
W3
W4
residence time (ms) volatile matter (% dmmf) fixed carbon (% dmmf) carbon (% daf) H/C (% dry) oxygen + sulfur (% daf)
0 24.4 47.5 83.7 0.65 9.9
100 21.8 47.0 78.0 0.65 15.7
200 12.0 51.6 82.3 0.45 12.4
300 3.7 57.0 91.2 0.19 55.4
400 2.7 60.9 91.7 0.14 44.8
0 35.9 33.6 84.0 0.81 8.2
120 20.5 65.8 84.7 0.49 99.7
160 23.0 62.0 84.4 0.53 99.8
200 10.0 75.8 88.3 0.38 66.7
240 4.3 74.4 90.6 0.17 66.5
cenospheric chars (as classified by Cloke and Lester9), while inertinite-derived chars were mainly honeycomb (dense) and unfused chars. Similar trends were found by Benfell et al.10 Image analysis was also used to measure the particle volume and pore size distribution. This showed that both the char porosity and pore size decrease in the order cenosphere > honeycomb > unfused char. The influence of increasing coal rank was also studied and found that the porous structure of inertinite-derived chars stayed relatively constant, while vitrinite-derived chars decreased in porosity. Consequently, at high ranks, the structural properties of chars generated from vitrinite and inertinite might behave similarly. There has been continuing efforts in the area of maceral separation by methods such as density gradient centrifugation, but these are time-consuming.12 It is mostly effective with demineralized samples and unrealistically small particle size for combustion studies.13,14 Thus, only limited work has been performed that concentrates on the combustion characteristics of single coal macerals, partly as a result of the limited availability of sufficient amounts of individual macerals.8,13 Therefore, these two maceral-diverse coals employed in this study offer the opportunity to investigate maceral influences during thermoplastic behavior and subsequent combustion reactivity for more size-appropriate coals. The behavioral differences are particularly interesting, because in comparison to northern hemisphere inertinite, the South African inertinite is more reactive, presumably as a result of the depositional environment and causation differences, and is termed “reactive semifusinite”.15 The effect of heat treatment on reactivity has been studied, and generally disordered carbons become less reactive as the heat-treatment temperature increases.16 Conversely, chars generated at higher heating rates are typically expected to achieve higher reaction rates because rapid heating can enhance thermoplasticity and porosity, which provides larger available surface areas.17 A drop-tube reactor (DTR) used in this investigation is a medium-scale device that is able to create an environment that more closely simulates industrial conditions by providing high temperatures, short residence times, and a high heating rate (∼105 °C/s). Because these operating conditions affect the char structure and, consequently, combustion reactivity, chars were generated in the DTR.
■
coal
sample name
to aid in the combustion of tars. Rather than true pyrolysis conditions, this 1% oxygen atmosphere is commonly used.17−19 Char samples were collected at different residence times by varying the height of the water-cooled sampling probe. Approximate residence times were calculated using gas flow rates, gas and bulk densities, particle size, and reactor specifications. For this estimation, it was assumed that all particles fall at the same velocity. Pre-screening experiments indicated that the vitrinite-rich coal would devolatilize more rapidly, and therefore, samples spanned the 240 and 400 ms residence times for the vitrinite- and inertinite-rich coals, respectively. Char Characterization. Proximate analyses were performed by ASTM D3172 for coal and char samples.20 The carbon, hydrogen, and nitrogen contents were determined using a LECO 600 CHN analyzer following ASTM D5373. A FEI Quanta 200 environmental scanning electron microscopy (SEM) was used to observe particle morphology and surface features. Petrographic analyses were performed with a reflecting light microscope to examine polished cross-sections of particles. Chars were embedded in an epoxy resin, left overnight to set, cut, and polished for the microscopic examination under oil immersion. To examine the optical texture, samples were observed under cross-polarized light. Char particle morphologies were identified on a point-counting basis and classified according to the system proposed by the International Committee for Coal and Organic Petrology.21 Surface area and porosity analyses were carried out using Micromeritics ASAP 2020. The samples were degassed for at least 24 h in vacuum prior to measurement. Nitrogen gas was used for analysis, at the analysis temperature of 78 K and a maximum manifold pressure of 925 mmHg. The adsorption measurements were translated to surface area based on the Brunauer−Emmett−Teller (BET) equation.22 The porosity distribution was calculated using the original density functional theory (DFT) considering slit pores. X-ray diffraction (XRD) patterns were obtained for char samples to determine the change in the size of the carbon crystallites during heat treatment. XRD samples were chemically treated with HCl/HF/ HCl, as suggested by Strydom et al., to remove ash and, hence, the disturbance of the ash on the spectra.23 Previous research has shown that this method of demineralization is the most effective without changing coal properties.50 XRD patterns were collected using a PANalytical X’Pert PRO MPD diffractrometer with Cu Kα radiation. To obtain the detailed pore size distributions, small-angle X-ray scattering (SAXS) was performed at beamline 12-ID-C at the Advanced Photon Source (APS) at Argonne National Laboratory. Char Combustion Reactivity. To evaluate reactivity, nonisothermal burnout profiles were determined using thermogravimetrical analysis (TGA).24 A char sample of 8 ± 0.1 mg was heated at a constant rate of 5 K min−1 with an airflow of 100 mL min−1. The burnout profile was used to determine temperatures of maximum mass loss rate and final burnout.
EXPERIMENTAL SECTION
■
Two South Africa coal samples were used: (1) run-of-mine Highveld coal with 88% inertinite maceral [dry mineral matter free (dmmf)], and (2) commercially cleaned (density separated) Waterberg coal with 92% (dmmf) vitrinite maceral. The coal samples were pulverized with a laboratory-scale ball mill and wet-sieved into a narrow size cut of 200 × 400 mesh (74 × 37 μm), which is similar to that of commercial pulverized coal combustion. Char Generation. Char samples were generated in a DTR heated to 1400 °C. An atmosphere of 1% oxygen in nitrogen was maintained
RESULTS AND DISCUSSION Char Morphology. Table 1 shows the proximate and ultimate analyses determined at different residence times for both coals at different stages of devolatization. Note that the original coals had similar carbon contents. The final char samples are almost fully devolatilized. The Waterberg (vitrinite113
DOI: 10.1021/acs.energyfuels.5b01517 Energy Fuels 2016, 30, 112−120
Article
Energy & Fuels
Figure 1. Inertinite-rich Highveld chars generated in DTR at (1a) 100 ms, (1b) 200 ms, (1c) 300 ms, and (1d) 400 ms and vitrinite-rich Waterberg chars generated at (2a) 120 ms, (2b) 160 ms, (2c) 200 ms, and (2d) 240 ms.
Figure 1), which was likely caused by the explosive ejection of built-up gases as explained by Gray.28 The open structure of these char particles will result in improved reactivity as a result of great access to oxygen. Higher magnification of chars at this residence time showed regions of smooth surface typologies. Figure 2a shows one of the many agglomerations of multiple particles of Waterberg chars at 240 ms. The adhesion of
rich) coal devolatilizes significantly faster than the Highveld coal (inertinite-rich). Figure 1 shows the morphology of drop-tube-generated chars, sampled at four different residence times. The SEM micrographs revealed differences in char morphologies, relating to differences in thermoplasticity of the two coals. After 100 ms (panel 1a of Figure 1), the Highveld coal still had mostly sharpedged particles. After 200 ms (panel 1b of Figure 1), the sharpedged particles transformed to slightly swollen particles with rounded edges. After 300 ms (panel 1c of Figure 1), a small number of pores as well as spherical particles had rough surfaces. Even though larger particles with visible pores and cracks were present after 400 ms (panel 1d of Figure 1), the majority were irregularly shaped “dense” chars lacking significant expansion as a result of a low degree of thermoplasticity of inertinite.25,26 Even at early stages of devolatization (panel 2a of Figure 1), sharp-edged Waterberg coal particles transformed to swollen spherical particles. At 120 ms, some spherical chars have smooth surfaces, while others have large bubble-derived pores (panel 2a of Figure 1). In panel 2b of Figure 1, after 160 ms, a larger particle size and larger surface pores were observed. According to Huang et al.,49 the presence of visible open surface “holes” (as seen in panel 2b of Figure 1) could be due to the expansion of existing bubbles or the localized acceleration of devolatization as a result of the catalytic effect of dispersed minerals. At 200 ms (panel 2c of Figure 1), bubble-derived pores beneath the particle surface are clearly visible, which implies a thin wall. Some smaller char fragments can be observed, probably caused by fragmentation of friable cenospheric char particles.27 Here, particles appear to reach a maximum size, after which they slightly decrease in diameter to form the wrinkled fabric particles shown in panel 2d of Figure 1. The decrease in diameter is likely due to the rapid release of volatiles within “balloon”-like gaseous spheres. Some particles have one “blow-hole” (as shown in panel 2c of
Figure 2. SEM micrographs of Waterberg chars.
particles could have been caused by plasticity of the char (caking) or the released tar, which acts as an adhesive and overwhelmed the available oxygen present in the system. Tetrahydrofuran (THF) can be used to extract the molecular weight fraction that includes tar.29 None of these agglomerations was found at higher residence times nor did THF extraction indicate the presence of tar (no solvent discoloration). According to Gray,28 the open structure of cenospheres with explosive ejection of the volatiles would allow for the heavy tar molecules to be ejected. Figure 2b is an example of a particle that experienced a high degree of thermoplasticity, allowing for extensive particle swelling (coal sample was
