Macromolecular Model of Coal from

Energy & Fuels 2002, 16, 791-801

791

Reexamining the Molecular/Macromolecular Model of Coal from Comparative in Situ Variable-Temperature 1H NMR Studies of Argonne Premium Coal 601 and Its Pyridine Extraction Residue Jincheng Xiong† and Gary E. Maciel* Department of Chemistry, Colorado State University, Fort Collins, Colorado 80526 Received October 25, 2001. Revised Manuscript Received December 19, 2001

We have carried out the first systematic in situ variable-temperature 1H NMR study, using the CRAMPS (combined rotation and multiple pulse spectroscopy) technique, of the residue from pyridine extraction of Argonne premium coal 601. The change of molecular mobility in the residue under various thermal treatments and pyridine swelling conditions were monitored quantitatively through a proton dipolar-dephasing experiment based on CRAMPS detection. Correlations among molecular structure, molecular mobility and solvent-extraction components were established through comparative variable-temperature 1H CRAMPS studies on the C5D5N-saturated original coal and its extraction residue. A critical review of the molecular/macromolecular (M/M) structural model of coal has been made on the basis of the new results, showing that the relationships between the M/M model and molecular dynamics is much more complicated than previously thought. Both the molecular phase and the macromolecular phase can show molecular dynamics behavior ranging from a lack of substantial motion to extremely mobile behavior with correlation times as short as 30 ns. Without systematic variable temperature studies, correlation between molecular mobility and the M/M model cannot be established correctly by NMR.

Introduction The molecular/macromolecular (M/M) model is a popular structural model of organic matter in coal.1,2 According to the M/M model, coal consists of a matrix based on an insoluble, three-dimensional, cross-linked network, called the “macromolecular phase”, and small organic molecules called the “molecular phase”, which are substantially soluble in nucleophilic solvents such as pyridine. The macromolecular phase is viewed as being composed of aromatic clusters that are cross linked by aliphatic bridging groups, and relatively flexible side chains covalently attached to the aromatic clusters. The molecular phase is considered to be either trapped in voids of the macromolecular matrix or attached to the macromolecular network via noncovalent associative forces, such as hydrogen bonds. One of the important advances in coal science in the past decade has been the realization that noncovalent associative forces make major contributions to stabilizing the structure of coal, including the macromolecular phase in coal.3-6 However, a, fundamental, molecular* To whom correspondence concerning this paper should be addressed. Phone: (970)491-6480. Fax: (970)491-1801. E-mail: maciel@ lamar.colostate.edu. † Current address: Dow Chemical Technical Center, 3200 Kanawha Turnpike, South Charleston, WV 25303. (1) Given, P. H.; Marzec, A.; Barton, W. A.; Lynch, L. J.; Gerstein, B. C. Fuel 1986, 65, 155. (2) Derbyshire, F.; Marzec, A.; Schulten, H.-R.; Wilson, M. A.; Davis, A.; Tekely, P.; Delpuech, J.-J.; Jurkiewicz, A.; Bronnimann, C. E.; Wind, R. A.; Maciel, G. E.; Narayan, R.; Bartle, K.; Snape, C. Fuel 1989, 68, 1091. (3) Gorbaty, M. L. Fuel 1994, 73, 1819. (4) Iino, M.; Takanoshashi, T.; Ohsuga, H.; Toda, K. Fuel 1988, 67, 1639.

level understanding of solvent effects and temperature effects on the noncovalent associative interactions in terms of the M/M model is still very limited. A detailed knowledge of such effects and further refining of the M/M model would open up possibilities for future coal conversion technology under mild conditions, rather than thermolysis. Although the M/M model of coal is widely accepted as a concept, the M/M model has been a subject of frequent debate in the literature,1,2 and the detailed nature of the molecular and macromolecular phases is still not fully understood. One of the best sets of evidence supporting the M/M model at a molecular level is from solid-state 1H NMR studies of original (untreated) coal and pyridine-saturated coal samples (actually samples saturated with perdeuterated pyridine, C5D5N).7-17 However, a direct correlation between mo(5) Nishioka, M. Fuel 1992, 71, 941. (6) Aida, T. In Proceedings, 45th Conference of Hokkaido Coal Research Group, 1989; p 11. (7) Jurkiewicz, A.; Marzec, A.; Idziak, S. Fuel 1981, 60, 1167. (8) Jurkiewicz, A.; Marzec, A.; Pislewski, N. Fuel 1982, 61, 647. (9) Barton, W. A.,; Lynch, L. J.; Webster, D. S. Fuel 1984, 63, 1262. (10) Jurkiewicz, A.; Bronnimann, C. E.; Maciel, G. E. Fuel 1989, 68, 872. (11) Jurkiewicz, A.; Bronnimann, C. E.; Maciel, G. E. Fuel 1990, 69, 804. (12) Jurkiewicz, A.; Bronnimann, C. E.; Maciel, G. E. In Magnetic Resonance of Carbonaceous Solids; Botto, R. E.; Sanada, Y., Eds.; Advances in Chemistry 229; American Chemical Society: Washington, D.C., 1993; p 401. (13) Jurkiewicz, A.; Bronnimann, C. E.; Maciel, G. E. Fuel 1994, 73, 823. (14) Xiong, J.; Maciel, G. E. Energy Fuels 1997, 11, 856. (15) Xiong, J.; Maciel, G. E. Energy Fuels 1997, 11, 866. (16) Xiong, J.; Maciel, G. E. Energy Fuels 2002. In press.

10.1021/ef0102566 CCC: $22.00 © 2002 American Chemical Society Published on Web 02/08/2002

792

Energy & Fuels, Vol. 16, No. 3, 2002

lecular mobility and the molecular phase has not been firmly established. There are still controversial issues among the results obtained using different NMR techniques and different coal-treatment conditions. The existence of a molecular phase is directly supported by solvent extraction studies of coals.1 Establishing a correspondence between the extractable fraction of coal and molecular mobility determined from NMR studies would be a major step in understanding the nature of the molecular and macromolecular phases and the function of noncovalent associative forces in coal. Although there have been many discussions and debates on these issues,1,2 these matters have not been settled. According to previous interpretations of the M/M model, the C5D5N-extraction residue would consist mainly of the macromolecular phase of coal. Comparative studies on the original coal and its extraction residue using proton dipolar-dephasing experiments should provide detailed information on the correlation of molecular mobility with both the macromolecular phase and the molecular phase in coal. Jurkiewicz et al. carried out such a comparative study at room temperature, using the proton dipolar-dephasing experiment with 1H CRAMPS (combined rotation and multiple pulse spectroscopy) detection.11 The 1H CRAMPS technique employed in their work was capable of resolving five different types of protons and the mobility of each proton type was monitored through their dipolardephasing behaviors. A rough correlation between mobility and extractible components was established in their work. However, in work described in a separate paper,16,17 we have shown that both untreated coal and C5D5N-saturated coal show very complicated molecular dynamics, which depends strongly on the temperature and action of the solvent, as well as the rank of coal. Without a systematic in situ variable-temperature NMR study of coal, a definitive correlation between molecular mobility and the extractible component cannot be established. Most previous NMR studies on coal samples resulting from solvent extraction were performed only at room temperature. However, solvent extraction is usually carried out near the solvent’s boiling point, so it is not surprising to see discrepancies generated from previous studies reported in the literature. In this work, we have for the first time carried out systematic, in situ, variable temperature 1H CRAMPS studies on the residue from pyridine extraction of an Argonne premium coal (601). With an improved dipolardephasing experiment based on 1H CRAMPS detection14,16,17 and a new procedure for data analysis,14-17 we have been able to monitor quantitatively the molecular dynamics of various structural moieties in the residue. Molecular mobilities in both the dried residue and the C5D5N-saturated residue were compared with those in the corresponding dry original coal and C5D5Nsaturated original coal, leading to correlations among molecular structure, molecular mobility and pyridineextraction components. These results provide the basis for critical review of the molecular/macromolecular (M/ M) structural model of coal, yielding a new level of understanding of the nature of the molecular phase and the macromolecular phase in coal and the relationships (17) Xiong, J. Ph.D. Dissertation, Colorado State University, Fort Collins, CO, 1996.

Xiong and Maciel

between the M/M structural model and molecular mobility, as detected experimentally by NMR. Experimental Section 1. Sample Preparation. The premium coal bank of the Argonne National Laboratory provided the untreated original coal 601 sample. The Argonne premium coal 601, from Blind Canyon, Utah (47.8% volatile material; 81.3% C, 5.4% H, 10.9% O, 1.6% N, O.4% S),14,16 is a high-volatile bituminous (HVB) coal and is substantially soluble in hot pyridine. Coal extraction was carried out for 21 days with refluxing pyridined5 (∼95 °C) in a Soxhlet extractor under a nitrogen atmosphere. The residue was dried in a vacuum oven (0.1 Torr) at 80 °C for 3 days, after which no further weight change was detected. A sample of the dried residue was packed and sealed into a homemade glass MAS rotor for variable-temperature 1H CRAMPS experiments between 25 and 180 °C. Details of the homemade MAS rotor were described in a previous paper.14 A sample of dried residue was also saturated with perdeuterated pyridine (C5D5N, Cambridge Isotope Laboratory, 99.94% 2H) with a weight ratio of 1:2 (one part coal:two parts pyridine). The sample was sealed in the homemade glass MAS rotor with epoxy resin. The sealed glass spinner with the pyridinesaturated coal sample was then put in an oven at 80 °C for 1 week to allow the establishment of equilibrium between pyridine and coal. 2. NMR Experiments. 1H CRAMPS (combined rotation and multiple pulse spectroscopy)18,19 experiments were performed on a severely modified Nicolet NT-200 NMR spectrometer, operating at a proton Larmor frequency of 187 MHz,17,18 using a Chemagnetics variable-temperature CRAMPS probe. The experimental setup and conditions of the 1H CRAMPS experiments were the same as those described for studying untreated and pyridine-saturated original coal samples in previous papers.14-17 Data analysis on proton dipolar-dephasing experiments was performed using an improved procedure and home-written programs, as described elsewhere.16,17

Results and Discussion 1. Changes in Chemical Functionalities of Coal Extraction Residue as Detected Using the 1H CRAMPS Technique. Knowing which chemical structures/functionalities in coal can be extracted in a solvent is not only valuable for understanding coal structure in general, but is also certainly a key for understanding interactions between the solvent and coal. Accordingly, we carried out an extraction of Argonne premium coal 601 with C5D5N and studied the resulting extraction residue. By comparing the 1H CRAMPS spectrum of the dried residue of C5D5N-extracted premium coal 601 (Figure 1a) with that of the original (untreated) coal (Figure 1b), one can see that the relative intensity of the aliphatic peak is lower for the extraction residue. This suggests that aliphatic protons are preferentially extracted by pyridine. This is in agreement with literature reports of both liquid-sample and solid-sample NMR studies on pyridine extracts,19 which show that the aliphatic component is preferentially populated in the pyridine extract. (18) Gerstein, B. C.; Chou, C.; Pembleton, R. G.; Wilson, R. C. J. Phys Chem. 1977, 66, 361. (19) Maciel, G. E.; Bronnimann, C. E.; Hawkins, B. L. In Advances in Magnetic Resonance: The Waugh Symposium; Warren, W. S., Ed.; Academic: San Diego, CA, 1990; Vol. 14, p 125.

Molecular/Macromolecular Model of Coal

Energy & Fuels, Vol. 16, No. 3, 2002 793

Figure 2. (a) 1H CRAMPS spectrum of C5D5N-saturated extraction residue of premium coal 601 obtained at 25 °C. (b) Simulated spectrum from computer peak deconvolution. In the simulated spectrum, dashed lines represent individual peaks from the deconvolution; the solid line represents the sum of all the deconvoluted peaks.

1H

Figure 1. CRAMPS spectra of premium coal 601 obtained at 25 °C: (a) dry residue from pyridine extraction, (b) untreated original coal, (c) C5D5N-saturated residue from pyridine extraction, and (d) C5D5N-saturated original coal. BR24 pulse sequence was used with a cycle time of 108 µs and a 90° pulse width of 1.2-1.3 µs. Each spectrum was acquired with 250 scans and a 3 s recycle delay. The MAS speed was between 1.5 and 1.8 kHz.

To obtain more detailed information on the changes of chemical functionalities implied in the above comparison, we took advantage of the enhanced resolution that results from pyridine swelling, as shown in Figure 1c and d. The 1H CRAMPS spectrum of the C5D5Nsaturated residue is similar to that of the pyridinesaturated original coal. One additional peak at about 4.9 ppm, with an intensity that is less than 1% of the total intensity, can be clearly seen in the 1H CRAMPS spectrum of pyridine-saturated residue (Figure 1c). This peak is assigned as water.11,12 To trace the origin of this peak, we checked the 1H NMR spectrum of the liquid C5D5N used for preparing the sample and found a small peak at 4.9 ppm, which could contribute partly to the 4.9 ppm peak observed for the C5D5N-saturated residue. A dry C5D5N sample (from a freshly opened ampule of another batch from Cambridge Isotopes Laboratory), for which the NMR spectrum is very clean at 4.9 ppm, was then used to prepare another sample of the C5D5Nsaturated residue. After the extraction residue was evacuated at 10-3 Torr for 48 h at 110 °C, and the dry pyridine was added to the evacuated dry residue and then sealed into the MAS rotor (with epoxy resin in a glovebox), the 1H CRAMPS experiment was repeated on the new sample; the 4.9 ppm peak was still visible with about one-third of the intensity observed previously. This result suggests that water in the dried extraction residue was released when the structure of the residue was swollen by pyridine. There are at least two explanations that come to mind for the fact that a distinct water peak was not observed before the pyridine was added (Figure 1a). First, the total water content is less than 1%. Second, the water molecules in the dried residue can be bonded strongly to the macromolecular structure of coal via hydrogen bonding, or can be intercalated into very small voids of the macromolecular matrix of coal; water molecules in such rigid environments would give a relatively broad

peak due to wide dispersion of chemical shifts and local magnetic susceptibility. Thus, the water peak in the 1H spectra of dried residue could be very weak and broad, so it would be difficult to detect. When pyridine is added to the residue, pyridine molecules can effectively break the cross links due to hydrogen bonds and cause water molecules to be detached from the macromolecular structure. Water molecules can also be released from very small voids due to significant swelling of the residue caused by pyridine saturation. Both mechanisms would significantly mobilize the residual water in the pyridine-saturated extraction residue of coal, thus producing a narrower H2O peak in the 1H NMR spectrum. Although the relative area of the water peak is still very low, the verticle intensity of the peak can be substantial because of the dramatic reduction of the peak width, which is why we can recognize the water peak easily in the 1H NMR spectra of pyridine saturated extraction residue of coal. This result provides a vivid example of the phenomenon in which small molecules that are trapped in small voids or associated with the macromolecular network via noncovalent interactions can show dynamical behavior that is similar to that of rigid macromolecules. Figure 2 shows the experimental 1H CRAMPS spectrum of the C5D5N-saturated residue from pyridine extraction of premium coal 601, and the simulated spectrum obtained by computer deconvolution. The chemical shifts, relative areas, and full-widths-at-halfheight (fwhhs) of the deconvoluted peaks are listed in Table 1. The assignments of these peaks are discussed in detail elsewhere.16,17 For comparison, the results of peak decovolution for the C5D5N-saturated original coal 601 are also listed in the same table. Several important differences between the original (untreated) premium coal 601 and its pyridine extraction residue can be seen from the Table 1. The 3.4 ppm contribution that has been identified in the 1H CRAMPS spectrum of the C5D5N-saturated original coal 60116 cannot be identified reasonably as a separate aliphatic peak in the C5D5N-saturated residue. Instead, the spectral region between 1.2 and 4.5 ppm in the spectrum of the pyridine extraction residue is better fit with a single peak at 1.7 ppm with a fwhh of 3.7 ppm, which is much larger than the fwhh (2.5 ppm) of the 1.7 ppm peak in the 1H CRAMPS spectrum of the C5D5N-saturated original coal. On the basis of

794


Xiong and Maciel

Table 1. Relative Concentrations (%) and fwhh of Various Types of Protons in the C5D5N-Saturated Premium Coal 601and Its Extraction Residue residue from pyridine extraction of premium coal 601 original premium coal 601a

a

chemical shift (ppm) fwhh (ppm) percentage (%) chemical shift (ppm) fwhh (ppm) percentage (%)

1.0 0.8 4.6 1.0 0.8 6.5

1.7 3.7 48 1.7 2.5 35

4.7 0.4 0.9 3.4 3.4 20

6.8 2.9 34 6.9 1.9 19

8.3 2.0 5.1 8.4 2.3 15

11.5 5.9 7.1 11.5 5.9 3.8

The values for the original premium coal 601 were taken from ref 16.

Figure 3. Stack plots of 1H variable-temperature CRAMPS spectra of the dried C5D5N-extraction residue of premium coal 601 obtained at (a) 25 °C, (b) 120 °C, and (c) 180 °C. The BR-24 pulse sequence was used with a cycle time of 108 µs and a 90° pulse width of 1.2-1.3 µs. Each spectrum was acquired with 200 scans and a 3 s recycle delay. The MAS speed was between 1.5 and 1.8 kHz.

dipolar-dephasing experiments discussed later in this paper, one finds that the aliphatic component in the extraction residue is less mobile than that in the original coal; i.e., pyridine saturation cannot mobilize the aliphatic structure in the residue as effectively as that in the original coal. That explains why a single broader peak between 1.2 and 4.5 ppm is observed in the 1H CRAMPS spectrum of the C5D5N-saturated residue. It is still possible that some of the protons with a chemical shift close to 3.4 ppm are in structural moieties that are preferentially extracted by pyridine. As shown in Table 1, the ratio of the fraction of aliphatic protons to that of the aromatic protons in the extraction residue is lower than that in the original coal. This is in agreement with the well-established fact that aliphatic structures are preferentially extracted.19 The intensity ratios of methyl protons (1.0 ppm) to nonmethyl aliphatic protons (1.7 and 3.4 ppm) in the original coal and its extraction residue are 0.13 and 0.080, respectively. This difference in the ratios suggests that structures containing methyl groups are preferentially extracted by pyridine. This implies that methyl groups in coal are mainly distributed in relatively small molecules, which can be easily extracted by pyridine. This view is consistent with the general trend that the average molecular weight in coal increases with C/H ratio.25 (20) Davis, M. F.; Quinting, G. R.; Bronnimann, C. E.; Maciel, G. E. Fuel 1989, 68, 763. (21) Tekely, P.; Nicole, D.; Brondeau, J.; Delpuech, J. J. J. Phys. Chem. 1986, 90, 5608.

2. Changes in Molecular Mobility of Coal Extraction Residue as Detected Using the in Situ Variable-Temperature 1H CRAMPS Techniques. Figure 3 shows stack plots of 1H CRAMPS spectra of the dried residue from C5D5N-extraction of premium coal 601 obtained using proton dipolar-dephasing experiments at temperatures of 25, 120, and 180 °C. To extract dipolar-dephasing constants from the dipolardephasing experiments, a procedure similar to what we discussed in our previous paper16 was employed here. The 1H CRAMPS spectrum of the dry residue obtained at each dipolar dephasing time was deconvoluted into aliphatic and aromatic peaks. Areas of the aliphatic and aromatic components were then used to construct experimental dipolar-dephasing curves, examples of which are shown in Figure 4. Each dipolar-dephasing curve can be simulated by a sum of two Gaussian decays, as described in the following eq 1.

M(t) ) MGF(0)e-t /(2T GF) + MGS(0)e-t /(2T GS) 2

2

2

2

(1)

The dipolar-dephasing time constants and fractions, MGF(MGF + MGS)-1 and MGS(MGF + MGS)-1, of both dephasing components obtained from the data analyses (22) Tekely, P.; Nocole, D.; Delpuech, J. J.; Julien, L.; Bertha, C. Energy Fuels 1987, 1, 1. (23) Kamienski, B.; Pruski, M.; Gerstein, B. C. Energy Fuels 1988, 1, 45. (24) Schulze, D.; Ernst, H.; Fenzke, D.; Meiler, W.; Pfeifer, H. J. Phys. Chem. 1990, 94, 3499. (25) Nishioka, M. Fuel 1991, 70, 1419.



Figure 4. Experimental proton dipolar-dephasing data (open circles) for the dried extraction residue of premium coal 601: (a) aliphatic proton at 25 °C, (b) aromatic proton at 25 °C, (c) aliphatic proton at 180 °C, (d) aromatic proton at 180 °C. The data were fitted with eq 1. Dashed curves represent individual dephasing components; the solid curve is the sum of all the dephasing components. Table 2. Dipolar-Dephasing Constants and Fractions of Fast and Slow Gaussian Dephasing Components in the Dried C5D5N-Extraction Residue of Premium Coal 601 fast Gaussian 25 °C 120 °C 180 °C

aliphatic aromatic aliphatic aromatic aliphatic aromatic

percentage

TGF (µs)

94 (89)a 100 (100) 89 (69) 91 (40) 80 (65) 77 (34)

10 (10) 15 (16) 10 (10) 16 (7.0) 11 (13) 13 (8.5)

slow Gaussian percentage

TGS (µs)

6 (11)

46 (33)

11 (31) 9 (60) 20 (35) 23 (66)

73 (59) 70 (48) 76 (62) 46 (44)

a The values in parentheses are corresponding numbers determined for original coal 601.14

are summarized in Table 2. For comparison, corresponding fractions and dipolar-dephasing constants of the original premium coal 601 are listed in parentheses.14 Similar to what we observed on the original coal,14 we did not see dramatic resolution changes in the 1H CRAMPS spectra of the dry residue at temperatures up to 180 °C, although the dephasing rates are reduced at high temperatures. The temperature dependence of molecular mobility, as reflected in dipolar-dephasing behavior, in the C5D5N-extraction residue is also qualitatively similar to that seen for the original coal in our previous work.14 As the temperature is increased, the fraction of the fast dephasing component decreases, and the fraction of the slow dephasing component increases. This suggests that rigid structures in the residue are thermally activated at higher temperature and change to more mobile states with limited anisotropic motion.14,16 It is no surprise to see that the fractions of slow dephasing components in the extraction residue at 120

and 180 °C are much less than those in the original coal. This observation is qualitatively in agreement with the common view that mobile species in coal should be relatively easy to remove via pyridine extraction. One sees in Table 2 that reduction of the fraction of mobile species in the residue measured at 120 and 180 °C is much more pronounced for aromatic protons than for aliphatic protons. For dipolar-dephasing measurements at 180 °C (Table 1), the slow dephasing aromatic component is reduced from 66% to 23% after extraction, while slow dephasing aliphatic component decreases from 35% to 20%. This seems to suggest that most of the mobile aromatic protons in the original coal are extracted by pyridine, while mobile aliphatic protons cannot be extracted as easy as mobile aromatic protons, a view that might initially seem to be in disagreement with our previous finding that aliphatic components are preferentially extracted. However, as seen in the discussion below, this kind of simplistic interpretation is often the basis for misunderstanding in the literature. Since molecular mobility in coal depends very much on thermal treatment and solvent swelling, the mobility measured at a particular condition cannot be simply used to interpret phenomena under different conditions. For example, thermal treatment alone cannot break hydrogen bonds in coal with temperatures up to 200 °C, while pyridine can effectively disrupt hydrogen bonds even at room temperature. Clearly, the structural moieties that can be mobilized easily by pyridine saturation would generally be different from the ones that can be mobilized under thermal treatment alone. Thus, there is no simple, direct correlation between mobile components in the original coal and extractable mobile components in the pyridine-saturated coal. Therefore, molecular mobilities determined for an untreated

796


coal or a dried extraction residue cannot be related simply and directly to either the solvent-extractable components of the coal or the molecular/macromolecular phases in coal. Tekely et al. studied the 1H dipolar-dephasing behavior of untreated coal indirectly through 13C CP-MAS detection at room temperature.21,22 They concluded that the mobile component in untreated coal contains relatively more aromatic hydrogen than does the immobile component. This result was discussed in several papers as an example that seems to contradict the chemical composition of the solvent extract.2,10,22,23 The results of Tekely et al. have been attributed to the inappropriate use of indirect detection of proton dipolar-dephasing behaviors through the 13C CP-MAS technique.24 However, as pointed out above, the relatively mobile protons detected in the untreated coal cannot be related simply and directly to the pyridine-extractable component in coal. Although indirect detection of proton behavior via 13C CP-MAS may underestimate the amount of mobile protons to a certain extent, this does not mean that the qualitative conclusions drawn by Tekely et al. are incorrect. Our 1H dipolar-dephasing results, based on direct 1H CRAMPS detection, clearly show that the relative fractions of aliphatic and aromatic protons in mobile components of an untreated coal depend on both the rank and the temperature of the coal sample.14-17 For untreated HVB coal at 180 °C, the fraction of aromatic protons in mobile components is higher than that of mobile aliphatic protons, as seen in Table 2. For untreated LVB coal, the fraction of aromatic protons in mobile components has been found to be lower than that of mobile aliphatic protons for temperatures up to at 180 °C. There is no simple correlation between molecular mobility in an untreated coal and the pyridine extractable components of the coal. To make more direct comparison of pyridine extractable components and molecular mobility, in the next section we discuss variable temperature 1H CRAMPS results obtained on the pyridine-saturated residue of premium coal 601. 3. Molecular Mobility in C5D5N-Saturated Extraction Residue of Premium Coal 601. Figure 5 displays 1H CRAMPS spectra of the C5D5N-saturated extraction residue of premium coal 601 obtained at 25, 60, and 90 °C. For comparison, the 1H CRAMPS spectrum of the dry residue obtained at 25 °C is also included in the Figure 5. No dramatic changes in the 1H CRAMPS spectra of the C D N-saturated residue are 5 5 observed when the temperature is raised from 25 to 90 °C. Small spectral resolution changes that occur with temperature variation are probably due to small changes in the spectral densities of random motion with correlation times around the cycle time of the multiple-pulse sequence (10-100 µs), which may interfere with coherent averaging of dipolar interactions by the multiplepulse sequence.14 Figure 6 presents stack plots of 1H CRAMPS spectra of the C5D5N-saturated residue obtained from dipolardephasing experiments at 25 °C and 90 °C. The dipolardephasing data were processed using a procedure similar to that described in a previous paper.14,16 As the overall resolution of 1H CRAMPS spectra of the C5D5Nsaturated residue is lower than that of the C5D5Nsaturated original coal, a reliable deconvolution between

Xiong and Maciel

Figure 5. 1H CRAMPS spectra of C5D5N-saturated extraction residue of premium coal 601 obtained at (b) 25 °C, (c) 60 °C, and (d) 90 °C. For comparison, the 1H CRAMPS spectrum of the dried residue is shown in (a). BR-24 pulse sequence was used with a cycle time of 108 µs and a 90° pulse width of 1.21.3 µs. Each spectrum was acquired with 250 scans and a 3 s recycle delay. The MAS speed was between 1.5 and 1.8 kHz.

the methyl peak (1.0 ppm) and the nonmethyl aliphatic component (1.7 ppm) could not be realized over the entire range of dephasing periods. Therefore, we combined the areas of these two peaks into the area assigned to aliphatic protons, and on this basis the quantitative analysis of the dipolar-dephasing behavior was performed in terms of aliphatic protons and aromatic protons in the C5D5N-saturated residue. Intensities derived from such deconvolutions of the experimental dipolar-dephasing spectra curves were fit for aliphatic and aromatic protons with a model described in the following equation:

M(t) ) MGF(0)e-t /(2T GF) + MGS(0)e-t /(2T GS) + 2

2

2

2

MLF(0)e-t/TLF + MLS(0)e-t/TLS (2) where MGF(0) and MGS(0) are the initial amplitudes of the fast and slow Gaussian dephasing components (GDC), respectively; MLF(0) and MLS(0) are the initial amplitudes of fast and slow Lorentzian dephasing components (LDC), respectively; TGF, TGS, TLF, and TLS represent corresponding dipolar-dephasing time constants. The fraction of each dephasing component can be calculated from its initial amplitude as

fi )

Mi(0) MGF(0) + MGS(0) + MLF(0) + MLS(0) where i ) GF, GS, LF, or LS (3)

Figure 7 illustrates experimentally derived dipolardephasing curves and the corresponding simulated curves. The dipolar-dephasing time constant and the fraction of each dephasing component obtained from the data analysis are summarized in Table 3. As shown in Table 3, the dipolar-dephasing curves for aliphatic protons in the C5D5N-saturated extraction residue of premium coal 601 consist of four compo-



Figure 6. Stack plots of 1H variable-temperature CRAMPS spectra of C5D5N-saturated extraction residue of premium coal 601 obtained at (a) 25 °C and (b) 90 °C. The BR-24 pulse sequence was used with a cycle time of 108 µs and a 90° pulse width of 1.2-1.3 µs. Each spectrum was acquired with 200 scans and a 3 s recycle delay. The MAS speed was between 1.5 and 1.8 kHz.

nents: fast Gaussian, slow Gaussian, fast Lorentzian, and slow Lorentzian. The mobile LDCs are produced by fast random molecular motion with correlation times in the range of 1 µs to 30 ns.16 It is interesting to note that there are still a total of 16% of mobile aliphatic LDCs left in the extraction residue at 25 °C. This clearly demonstrates that not all of the very mobile components, with Lorentzian proton dephasing characteristic, can be extracted. However, compared with the pyridinesaturated original coal,16 the fractions of fast and slow LDCs of aliphatic protons in the C5D5N-saturated residue are dramatically decreased as a result of pyridine extraction. At 90 °C, fLF and fLS of the aliphatic protons in the C5D5N-saturated original premium coal 601 are 33% and 31%, respectively. At the same temperature, these two fractions in the C5D5N-saturated residue are reduced to 13% and 7%, respectively. Clearly, the slow LDC is preferentially extracted, presumably because a major fraction of slow LDC is made up of highly mobile, small molecules. When the temperature is increased from 25 °C to 90 °C, about 8% of the fast aliphatic GDC change to more mobile states in the C5D5N-saturated residue. With the same temperature increase, 29% of the fast aliphatic

GDC change to LDCs in the C5D5N-saturated original coal. Thus, the aliphatic structures left in the C5D5Nsaturated extraction residue are more rigid than those in the C5D5N-saturated original coal 601. This result also suggests that a significant fraction of the small molecules present in coal are buried in a rigid macromolecular matrix, even in the pyridine-saturated coal at 25 °C, and they can be released with an increase of temperature. As most such small molecules are transferred to the extract during extraction, there is a much smaller amount of small molecules left in the rigid macromolecular matrix of the residue than was present in the original coal. This also explains why a much smaller fractional change of the fast Gaussian component of aliphatic protons is observed with an increase in temperature in the C5D5N-saturated residue than is the case in the C5D5N-saturated original coal 601. Just as in the case for aromatic protons in the C5D5Nsaturated original coal 601, the dipolar-dephasing curves of aromatic protons in the C5D5N-saturated extraction residue can be modeled with three dephasing components: fast Gaussian, fast Lorentzian, and slow Lorentzian. As expected, the fLS value of aromatic protons in the C5D5N-saturated extraction residue is lower than

798


Xiong and Maciel Table 4. Fractions (%) of Various Types of Protons in the C5D5N-Saturated Extraction Residue of Premium Coal 601a 25 °C 90 °C

aliphatic aromatic aliphatic aromatic

FGF (%)

FGS (%)

FLF (%)

FLS (%)

36 29 32 21

13

6.4 11 7.5 20

2.7 2.4 4.2 1.9

14

a F , F , F , and F GF GS LF LS are fractions of the fast GDC, slow GDC, fast LDC, and slow LDC, respectively, relative to the total number of protons in the residue (100%).

Table 5. Fractions (%) of Various Types of Protons in the C5D5N-Saturated Premium Coal 601a 25 °C 90 °C


FGF (%)

FGS (%)

FLF (%)

FLS (%)

36 20 20 14

7.1

9.1 12 21 17

12 4.6 20 4.6

3.5 -

a F , F , F , and F GF GS LF LS are fractions of the fast GDC, slow GDC, fast LDC, and slow LDC, respectively, relative to the total number of protons in premium coal 601.

Figure 7. Experimental dipolar-dephasing data (open circles) of (a) aliphatic and (b) aromatic protons in the C5D5Nsaturated extraction residue of premium coal 601 obtained at 90 °C. The data were fitted with eq 4.1. Dashed curves represent individual dephasing components; the solid curve is the sum of all the dephasing components. Table 3. Dipolar-Dephasing Time Constants and Fractions of Various Types of Protons in the C5D5N-Saturated Extraction Residue of Premium Coal 601 25 °C fGF (%)a TGF (ms)b fGS (%)a TGS (ms)b fLF (%)a TLF (ms)b fLS (%)a TLS (ms)b

90 °C

aliphatic

aromatic

aliphatic

aromatic

63 (56)c 0.011 22 (11) 0.048 11 (14) 0.20 4.7 (19) 4.5

69 (55) 0.019

55 (31) 0.012 25 (5.0) 0.061 13 (33) 0.30 7.2 (31) 5.6

49 (38) .020

26 (33) 0.13 5.6 (13) 8.1

47 (49) 0.10 4.4 (13) 11

a The fractions (%) of the four dephasing components are denoted as fGF, fGS, fLF, and fLS. b TGF, TGS, TLF, and TLS denote the dipolar dephasing time constants of fast Gaussian, slow Gaussian, fast Lorentzian, and slow Lorentzian dephasing components, respectively. c The values in parentheses are corresponding fractions in C5D5N-saturated original premium coal 601.16

fLS of the C5D5N-saturated coal. But the differences between the fractions of the three dipolar-dephasing components of aromatic protons in the extraction residue and the original coal at 90 °C are much less pronounced than those for aliphatic protons. This suggests that aromatic structures are not changed as significantly as aliphatic structures by pyridine swelling/extraction. In particular, the fLF value measured in the extraction residue at 90 °C is essentially unchanged by pyridine extraction. Thus, it appears that the fast Lorentzian dephasing component of aromatic protons is essentially nonextractable. This component may be due to aromatic structures in macromolecular chains that undergo mobile segmental motion. It is also possible that this component corresponds to aromatic structures in flexible side chains that are attached to the macromolecular network through single covalent bonds. We will see later that the former possibility is more plausible.

Table 6. Fractions (%) of Various Types of Protons in the C5D5N-Saturated Extracted Residue on the Basis of the Total Number of Protons in the Original Coal (100%), as Scaled by the Extraction Yield of 40% 25 °C 90 °C


FGF (%)

FGS (%)

FLF (%)

FLS (%)

22 18 19 12

7.6

3.8 6.6 4.5 14

1.6 1.4 2.5 1.1

8.7

4. Correspondence between Extractable Components and Molecular Mobilities. To unveil the differences between the original coal and its pyridine extraction residue, a comparison of the fractional changes of different components based on the total number of aliphatic and aromatic protons with different dipolardephasing characteristics in the sample should be informative. As shown in Tables 4 and 5, the relative fractions of each component listed in Table 3 of this paper and Table 3 in ref 16 were converted to fractions relative to the total number of aliphatic and aromatic protons in the C5D5N-saturated extraction residue and in the C5D5N-saturated original coal, respectively. To compare the number of protons of each component between Tables 4 and 5, one needs to set a common base. The fractions in Table 4 can be scaled down according to the extraction yield, using the total number of protons in the original coal as a common base (100%). Nishioka studied the pyridine extraction yield of premium coals of various ranks.25 The yields of extracts from HVB coals were determined to be from 37 to 44 wt % on the dried ash-free (daf) basis. By comparing the difference in weight of the dry original coal 601 and its dried pyridine extraction residue, we determined that the extraction yield of our extraction procedure was 40%. This is the value that was used to generate the numbers in Table 6, which lists the fractions of each component in C5D5N-saturated residue on the basis of the total number of protons in the original coal. A correlation between molecular mobility and extractable components can be established directly by comparing Tables 5 and 6. As 90 °C is very close to the Soxhlet extraction temperature that we used (∼95 °C), fractional changes of the various categories of protons at 90 °C


should be more meaningful for interpreting our extraction results than would be data obtained at some other temperature. Considering the experimental errors involved in the fractions and uncertainties in the extraction yield, we will not emphasize small changes in fractions in the following discussion. Fast Gaussian dephasing components in coals correspond to very rigid structures. The slow Gaussian dephasing component is due to rather rigid structures with very limited anisotropic motion. Previously, Gaussian dephasing components have always been assigned in the literature as protons in the rigid macromolecular phase of coal and have been regarded as unextractable components. The results presented in Tables 5 and 6 clearly show that this interpretation is not entirely correct. In the C5D5N-saturated premium coal 601 at 25 °C, 43% of the protons are aliphatic protons with Gaussian dephasing characteristics, part of a total aliphatic proton content of 64% (Table 5). After pyridine extraction, only 30% of the original total proton content remains as aliphatic protons with Gaussian dephasing behavior, part of a total 35% of the original protons remaining in the residue as aliphatic protons. This clearly shows that a substantial fraction of the aliphatic protons that show Gaussian behavior at 25 °C in the original coal are extracted, supporting our conclusion that a significant quantity of small molecules is buried in tight voids of the macromolecular matrix or attached to the macromolecular network via hydrogen bonds, even in the pyridine-saturated coal at 25 °C. When the temperature of a pyridine-saturated coal is increased from 25 °C to 90 °C, thermally more energetic pyridine molecules can break a larger number of, and stronger, hydrogen bonds and enlarge the small voids in the rigid macromolecular matrix. This leads to a further release of small molecules that had been either attached to or trapped in the rigid macromolecular structures at 25 °C. A pronounced increase of the fraction of LDC in the C5D5N-saturated premium coal 601 at 90 °C can be seen in Table 5, especially for aliphatic components. One can also see from Tables 5 and 6 that the number of protons showing Gaussian dephasing behaviors in the C5D5N-saturated coal at 90 °C is close to that in the C5D5N-saturated residue at 90 °C. Considering the fact that 90 °C is close to the extraction temperature used in this work, this result suggests that almost all of the small molecules that can be mobilized by pyridine under extraction conditions are already mobilized in the C5D5N-saturated coal at 90 °C. It is well-known that the pyridine extraction process is very slow and that refluxing with frequently refreshed solvent is needed to achieve high yields. On the basis of the above results, it seems that mobilization of small molecules is not a kinetic bottleneck in the extraction. The extraction rate seems to be mainly determined by molecular diffusion from enlarged pores to the liquid phase. The number average and weight average of molecular weight of the pyridine extract of HVB coals were determined to be on the order of 600 and 1600, respectively, using size exclusion chromatography.2 So, the typical sizes of extracted molecules are much larger than that of pyridine molecules. Pyridine could easily diffuse into rigid structures, break hydrogen bonds around small


molecules, and create large pores inside the rigid macromolecular matrix. However, the partially mobilized molecules may not be able to diffuse out of the pores as easily as pyridine diffuses in. In this view, the extractability of coal depends not only on the sizes of pores, and molecules to be extracted, but also on the connectivity of pores created by pyridine swelling. As changes in the dipolar-dephasing time constant reflect mainly changes in the rotational mobility of a molecule, the dephasing constant depends mainly on the size of the pore in which the molecule is located, and is not very sensitive to the connectivity of pores. Connectivity could be a factor that plays a role in determining the amount of LDC that is not extractable. As can be seen from Tables 5 and 6, most of the slow LDCs, for both aliphatic and aromatic components, are extracted by pyridine. This implies that large pores tend to have good pore connectivity in the C5D5N-saturated coal, suggesting that in the original premium coal 601 exist large domains that are relatively rich in hydrogen bonding bridges, so that large connected pore systems can be created with pyridine saturation. The unextractable portion of the slow LDC could be due to small molecules trapped in large, closed voids or very flexible side chains attached to the macromolecular network. Most of the aliphatic protons in the C5D5N-saturated premium coal 601 with fast Lorentzian dephasing characteristics are also extracted by pyridine, but to a smaller extent than those with slow Lorentzian dephasing characteristics. Comparing 90 °C data in Tables 5 and 6, we see that about 82% of the combination of fast and slow aliphatic LDC and slow aromatic LDC is extracted. In contrast, 82% of the fast aromatic LDC at 90 °C remains in the residue. Thus, the fast LDC of aromatic protons is not effectively extracted, suggesting that this component is likely due to mobile aromatic structures with segmental motion. This suggestion is based on the fact that the extraction residue can still be swollen effectively by pyridine, yielding a material with rubber-like behavior, which implies that the macromolecular chain segments are still very mobile. Segmental motion of the mobile macromolecular chains is most probably manifested in dipolar-dephasing experiments as aromatic protons with fast Lorentzian dephasing components. As discussed in another paper16 and shown in Table 5, this assignment is also consistent with the fact that the fast Lorentzian dephasing component of aromatic protons in the C5D5N-saturated coal 601 cannot be converted further to slow Lorentzian components when the temperature is increased from 25 to 90 °C. In contrast to C5D5N-saturated premium coal 601, there is no identifiable fast LDC in C5D5N-saturated premium coal 501, a low-volatile bituminous coal (Pocahontas No. 3; 19.0% volatile matter, 91.8% C, 4.5% H, 1.7% O, 1.3% N, 0.5% S).14,16 This difference between the two coals is related to fundamental structural differences between HVB and LVB coals. The macromolecular structure of a HVB coal is believed to consist of relatively small aromatic units connected by flexible aliphatic cross links.25,26 Hydrogen bonding bridges contribute to stabilization of the macromolecular net(26) Schobert, H. H. The Chemistry of Hydrocarbon Fuels; Butterworths: London, 1990.

800


work and make it rather rigid. When the hydrogen bonding bridges are broken by pyridine, the aromatic structure in the macromolecular chain could gain substantial segmental mobility in HVB coals. In LVB coals, such as premium coal 501, the aromatic units are much larger, and connected by shorter and less flexible cross links; this makes the aromatic structure much more rigid, so segmental motion of aromatic units in the LVB coal cannot be enhanced effectively by pyridine. It should be noted that Painter and co-workers studied pyridine swelling of coals between 25 and 100 °C,27,28 finding minimal variation of swelling over this temperature range. They viewed their results within the theoretical framework of polymer physics and concluded that those results were consistent with the model of coal as a macromolecular network. 5. The M/M Model of Coal and Molecular Mobility. Our variable temperature studies on both untreated coal and C5D5N-saturated coal support the basic idea of the M/M model. However, the relationships between the M/M model and molecular dynamics is much more complicated than previously thought. A significant fraction of the very rigid component is always detected in our 1H CRAMPS experiments on coal, no matter whether the coal sample is subjected to thermal treatment or pyridine saturation. This rigid component is typically attributed to the existence of a highly cross-linked macromolecular structure; however, this rigid component is not equivalent to the macromolecular component in coal. In this work, we have shown that small molecules can be buried in tight voids of the macromolecular matrix or attached to the macromolecular network of coal and show dynamical behavior that is similar to that of rigid macromolecules, even when the coal is saturated with pyridine. This work also suggests that the macromolecular network of coal can undergo fast segmental motion under pyridine saturation, and thus show Lorentzian dephasing behavior with dephasing time constants as long as 0.3 ms. The molecular phase in coal shows even more complicated dynamical behavior than exhibited by the macromolecular phase. Our work demonstrates that extractable small molecules in coal can, in principle, exhibit a wide variety of molecular dynamical behavior, depending on the structure and surrounding matrix of the molecule and the sample temperature. Side chains covalently attached to the macromolecular network in coal can also show rotational mobility that is similar to that of small molecules in coal. In principle, small, mobile molecules would have much higher translational mobility than flexible side chains. However, since proton dipolar-dephasing behavior is actually determined by rotational mobility instead of translational mobility, the dipolar-dephasing experiment alone is not sufficient for us to differentiate the molecular phase from flexible side chains. For example, unextractable, slow LDC components could be due to very flexible side chains or small molecules confined in a closed pore system. As Given et al. have suggested,1 diffusion measurements could be potentially useful for separating contributions from small molecules and flexible side chains. (27) Cody, G. D.; Eser, S.; Hatcher, P.; Davis, A.; Sobkowiak, M.; Shenoy, S.; Painter, P. C. Energy Fuels 1992, 6, 716. (28) Cody, G. D.; Painter, P. C. Energy Fuels 1997, 11, 1044.

Xiong and Maciel

A problem that has been discussed many times in the literature is whether the mobile components in coal detected by NMR overestimate the fraction of a molecular phase.2 However, our results show that there is no simple one-to-one correspondence between the molecular phase and molecular dynamics. Although a large fraction of the mobile LDCs are small molecules, both flexible macromolecular side chains and macromolecular main-chain structures undergoing fast segmental motion make a significant contribution to mobile LDCs. This seems to suggest that the mobile components detected by NMR overestimate the molecular phase in coal. Moreover, we have also shown that a significant fraction of the small molecules in what most researchers would consider the molecular phase behave as rigid components in coal. Without a systematic, variabletemperature study of coal under a variety of conditions, meaningful delineations cannot be made. And, it is unlikely that there will be a universal answer for coals of different ranks and origins. As small molecules can be buried in tight voids of the macromolecular matrix of coal, some of these molecules may not be mobilized without breaking covalent bonds of the macromolecular structure. Once the covalent bonds are broken, there will be a very small chance of distinguishing the small molecules that are buried inside the coal matrix of an untreated sample from small molecules that are generated by the breakdown of the macromolecular network. In this regard, the conceptual molecular phase may never be completely detected experimentally. Conclusions In this work, we have demonstrated that in situ variable-temperature 1H NMR studies, based on CRAMPS detection, are very powerful for elucidating dynamical and structural details of pyridine/coal interactions. With comparative studies on a C5D5N-saturated coal and its residue, a correlation among pyridine extractable components, molecular mobility, molecular structure and the M/M model have been established. The following conclusions can be drawn from this study: Molecular mobilities determined for an untreated coal or a dried extraction residue cannot be related directly to either the solvent-extractable components of the coal or the molecular/macromolecular phases in coal. Molecular mobility in coal depends very strongly on temperature. Without systematic variable-temperature studies, correlation between molecular mobility and extractable components cannot be correctly established by NMR. NMR experiments performed at the extraction temperature provide a more direct correlation between molecular mobility and extractable components than do comparisons of room-temperature results. Gaussian dephasing components (dephasing time constants: 10-80 µs) determined from 1H NMR measurements on a C5D5N-saturated coal can contain significant contributions due to small molecules buried in or attached via noncovalent associations to the macromolecular matrix. Such small molecules can be extracted by pyridine. Most of the aliphatic protons with Lorentzian dephasing characteristics (dephasing time constants: 0.1310 ms) in C5D5N-saturated premium coal 601 at 90 °C


can be extracted by pyridine, thus belonging to the molecular phase. A small but significant fraction of very mobile Lorentzian dephasing components cannot be extracted by pyridine. They can be due to very flexible macromolecular side chains or small molecules confined in a large, closed pore system in the coal. The aromatic proton component showing fast Lorentzian dephasing behavior (time constants: 0.2-0.3 ms) is largely unextractable. This dephasing component may correspond to aromatic structures in macromolecular chains that undergo fast segmental motion. Aliphatic components in premium coal 601 are preferentially removed during pyridine extraction.


Mobilizing the molecular phase is not a limiting step in pyridine extraction of coal. Instead, pyridine extraction efficiency is limited by diffusion of small molecules in the macromolecular matrix of coal. These new experimental results clearly demonstrate that relationships between the M/M model and molecular dynamics is much more complicated than previously thought. Nevertheless, the M/M model is still very useful as a working hypothesis and for stimulating further studies on the complex structure of coal. Acknowledgment. This work was supported by DOE grant DE-FG22-93PC93206. EF0102566

Macromolecular Model of Coal from

Recommend Documents