FUEL

PROCESSING
TECHNOLOGY
ELSEVIER Fuel Processing Technology 7 7 -7 8 (2002) 2 5 -3 2
www. elsevier. com/locate/fuproc

Towards an understanding of the coal structure:

Received 31 January 2002; accepted 20 March 2002

In memoriam Professor Dr. Peter H. Given in gratitude and admiration for his scientific activity.

Abstract

This review is limited to studies that directly refer to the following concepts: (i) macromolecular
structure of coal; (ii) the contribution of relatively low molecular substances in the structure; (iii)
factors effecting capabilities of solvents in coal extraction and the extraction mechanism; as well as
(iv) intermolecular forces occurring in coal organic matter. Although concept (i) was introduced in
the 1970s, no great progress has been achieved. Available data for a few coals indicate that mass
distribution of the macromolecular part are in the 1300-3500 amu range. There are no experimental
data available that could describe the structure of the macromolecular part of coal. Referring to item
(ii), the presence of significant amount of various substances in the ~ 100-500 amu range in coals
was evidenced by mass spectrometry methods (field ionization mass spectrometry (FIMS) and
matrix-assisted laser desorption/ionization-time-of-flight (MALDI-TOF)). As for item (iii), major
contributors that influence solvent extraction capability is the electron-donicity of the solvent as well
as the shape and size of its molecules. One major step of the mechanism of coal extraction is the
substitution phenomenon. Presumably (in reference to item (iv)), all known intermolecular forces
hold together coal organic matter. However, only hydrogen bonds can be directly detected, as yet.
© 2002 Elsevier Science B.V. All rights reserved.

Keywords: Coal; Macromolecules; Molecules; Extraction mechanism; Intermolecular forces; FIMS; MALDI

1. Introduction

An extensive research on coal structure started in the 1970s, shortly after the World Oil
Crisis. One of the alternatives was to get knowledge that could contribute to developing

E-mail address: marzec@karboch.gliwice.pl (A. Marzec).

0378-3820/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 -3 8 2 0 ( 0 2 ) 0 0 0 4 5 - 0
26 A. Marzec / Fuel Processing Technology 77 -7 8 (2002) 25-32

new economic technologies for the production of petroleum substitutes from coal. Another
factor was the availability of new instrumental and analytical techniques. The present
knowledge of coal structure has been a result of investigation carried out by hundreds of
researchers. Therefore, it is impossible to describe their efforts and results in a short
review. This paper is rather aimed at supplementing the already published reviews [1-6].

2. Average molecule concept of coal structure

Since 1960s, coal chemists attempted to create models of coal structure which would
represent a synthetic view of principle characteristics of coal organic matter. One of the
strategies consisted of constructing “average structural unit” of coal organic matter.
Structural units were, in fact, molecules described in a classically chemical way using
atoms, chemical bonds and some functional groups. Probably, Fuchs and Sandhoff worked
out the first such model in 1942. Later, other average structural units were published by: A.
Gillet in 1950; L. Cartz and P.B. Hirsch in 1960; P.H. Given in 1960; W.R. Ladner in 1963;
W.H. Wiser in 1975; G.J. Pitt in 1979; L.A. Heredy and I. Wenderin 1980; C. Spiro in 1982;
and J.H. Shinn in 1984 (for drawings of the models, see Refs. [1,3-5]). The structures had
been built on the basis that they were characterized by the same values of parameters (such as
elemental composition; carbon and hydrogen aromaticities; number of rings; size of
aromatic clusters, etc.) as were determined for the total organic matter of some coals or
for its major petrographic component i.e., vitrinite. However, in 1984, P.H. Given questioned
the view that one or even several molecules can be a representative of a coal organic matter
[6a]. Hence, average structural models are not very useful and they rather represented some
sort of professional divergence among chemists when they were involved in coal research.
In the late 1970s, coal scientists began to apply a number of physical methods suitable
for solid-state characterization. One of the methods was transmission electron microscopy
(TEM) in its dark field (DF) mode. A. Oberlin and her collaborator J.N. Rouzaud
developed it [7,8]. The technique enabled researchers to directly observe, in coal organic
matter, the diameters of polyaromafic units (BSU) as well as their space orientation. It was
shown [7,8] that the sizes of BSU are around 1 nm (which may correspond to the range of
4 -1 2 condensed aromatic rings). For bituminous coals, the sizes remain nearly constant
up to the anthracite range. In lower-rank coals, the BSUs are randomly oriented and are
held so by the presence of various, mostly heteroatomic, chemical bonds. For higher-rank
coals, increasing parallel orientation of BSUs is observed. The TEM-DF technique is not
able to observe objects smaller than 0.7 nm.

3. Macromolecular concept of coal structure

In the meantime, a new idea began to penetrate coal science due to the influence of
polymer science. Most man-made polymers were known to be soluble in common organic
solvents in spite of their extremely high molecular weights. The insoluble exceptions were
the cross-linked polymers that formed three-dimensional network. Since coals were known
to be rather insoluble, a view appeared that coal organic matter might show such cross-linked
 A. Marzec / Fuel Processing Technology 7 7 -78 (2002) 25-32 21

polymer nature. Having in mind this view, Larsen and Kovac [9] used the statistical theory of
polymer network to calculate the average molecular weight (Mc) of a chain (composed of
aromatic and hydroaromatic units linked together) between two cross-links in various coals,
from experimental data on solvent swelling and mechanical properties of coal. They
concluded that Mc for bituminous coals was in the 1500-1800 amu range. Later, a great
deal of attention was paid to the concept of cross-linked polymer character of coals.
Measurements of Mc were continued by Larsen and Kovac as well as by Lucht and Peppas
[10] and a number of others. The problem was extensively reviewed by Peter Given [6b]. He
concluded: “we must accept that coals are cross-linked macromolecular solids. B ut.. .they
are sui generis, of a class by themselves, of unrivalled complexity. ” Recently, the macro-
molecular character of at least, part of coal organic matter, was also experimentally
confirmed by Krzesinska [11] who applied molecular acoustics methods. The methods,
however, cannot discriminate between cross-linked and entanglement structures.
The problem of occurrence and chemical character of cross-links has been not clarified
as yet. An analogy between insoluble polymers and insoluble part of coal implied that coal
cross-links are formed by covalently linked atoms or groupings (ether, methylene chain,
sulfide, etc). However, coal extraction experiments (carried out in mild conditions that
most likely exclude a rupture of covalent bonds) clearly show that the content of coal
insoluble part depends on the solvent used for extraction. For example, studies on
extraction with the use of binary and ternary solvent systems—pioneered by lino et al.
[12] and recently developed by Dyrkacz and co-worker [13,14]—indicated that the coal
insoluble part assumed to be covalently cross-linked matter, might constitute less than 30
wt.% (data for 87% Cdaf coal: 70 wt.% of the coal was solubilized by CS2 /NMP mixture).
The same coal contains approximately 65 wt.% of insolubles when neat NMP solvent is
used. The question is whether there are any cross-links in the part of coal matter that is not
soluble in NMP but can be dissolved when CS2 /NMP is used. And what is the content of
the covalent three-dimensional network that cannot be dissolved in any organic solvent?
There is also a lack of analytical data referring to the composition of high yield extracts
resulting from the application of binary or ternary solvent mixtures although there are
available analytical techniques (LD or MALDI-TOF mass spectrometry; for explanation
of the acronyms, see Section 4.2) that could be used for tracing low molecular as well as
very high molecular weights. Such available data for a few raw coals indicate that mass
distribution of macromolecules are in the 1300-3500 amu range.
Recently, the concept of network structure of coals and nature of forces that bind the
network together had been discussed by lino [15]. The conclusion is “there is no clear
evidence for covalent network model, though it is sure that (such) network exists in coal
from swelling and elasticity measurements of coals.”

4. Molecular components of coal structure

4.1. Coal extraction studies

For long time, not much attention had been paid to the soluble part of coals and its role
as the component of coal structure. The data on coal extraction referred to the application
28 A. Marzec / Fuel Processing Technology 77 -7 8 (2002) 25-32

of organic solvents at their boiling points or at higher temperatures when extractions were
carried out in autoclaves [6c]. Well before the 1980s, it was already known that low
molecular substances are components of coal organic material. However, the following
questions remained unclear: what is the content of the low molecular substances in coal;
what is their chemical nature and molecular weight distribution; why extraction capa­
bilities of various organic solvents are so different; and what forces hold together the low
molecular and macromolecular substances.
In 1977, systematic studies of coal extraction had been undertaken. First of all, with the
aim to avoid cleavage of covalent bonds, extractions were carried out at room temperature
and the extracts were freed from solvents in rotary evaporators under reduced pressure
[16-18]. Numerous organic solvents used for coal extraction had been characterized by
quantitative measures [16] of their electron-donor (DN) and electron-acceptor (AN)
properties (Gutmann’s [19] donor and acceptor numbers) as well as by Hildebrand’s
solubility parameter and Snyder’s polarity index. Correlations were found between the
extract yields on one hand and donor numbers of the solvents as well as the differences
(DN —AN), on the other. Later studies revealed much stronger correlations in question
when a logarithmic coordinate was used [20]. Thus, there is no question that nucleophilic
solvent properties and, to a lesser extent, electrophilic properties, are determinant factors
of solvent extraction capabilities. In general terms, the mechanism of coal extraction seems
to consist in substitution: electron-donating solvent molecules replace the coal electron-
donor participants in the forces [16]. The higher the electron-donor capability of the
solvents, the more coal electron-donor centers are replaced by solvent molecules and
therefore, the higher the extract yields [16]. Additional but significant information on the
extraction mechanism arose from the detailed studies of various binary/ternary solvents
[14]. The authors concluded: “the basicity of the solvent is a major contributor to the
extraction of a coal” however, “two other factors may also be acting.” One is a steric
requirement of the flat molecular shape (characteristic for aromatic rings compared with
saturated rings). Flat molecules likely have better access to coal matter interior. The second
one is unsaturation of a fragment of solvent molecule that may lead to P i-P i interactions
with coal. The interactions presumably disturb original P i-P i forces in coal.

4.2. Mass spectrometry (FIMS, MALDI) o f coal extracts and o f raw coals (MALDI)

Molecular masses of the extracts components were determined [17,18,21] with the use
of FIMS (field ionization mass spectrometry), the technique that does not produce
fragmentation ions. Moreover, successive heating of the extract samples inside the mass
spectrometer (and the successive mass spectra recording) enabled investigators to get
information on the extract composition without any interference with the sample thermal
decomposition in the mass spectrometer. The detailed description of the applied FIMS
technique can be found elsewhere [21,22]. The chemical nature of the components was
determined by means of high-resolution MS [17,18,21] and low-voltage electron impact
MS [23]. It is worth pointing out that FIMS spectra of the coal extracts (as well as of the
extraction residues) showed extremely complex composition of coal organic matter. The
estimated number of the detected extract components was approximately 2000 individual
substances [21]. The number might be much higher if isomers could be detected by FIMS.
 A. Marzec / Fuel Processing Technology 7 7 -78 (2002) 25-32 29

Summing up, the results for 81% Cdaf bituminous coal were as follows. The
molecular mass distributions were similar, no matter what solvent was used nor what
was the extract yield: the extracts were composed of substances in the 70-800 amu
range, but substances in the 200-500 amu definitely predominated [17,18,21]. These
substances constituted approximately 30 wt.% of the bituminous coal. The identified
coal molecules represented the following classes: C„H2„ r (z - 6-28); CHN, CHN2,
CHN3; CHO, C H 02, C H 03; CHON, CHON2, CH N 02, CHN20 2, CHN30 2; CHS,
CHNS, CHOS [17,18,21,23],
In 1995, laser desorption (LDMS) and matrix-assisted laser desorption/ionization
(MALDI) coupled with time-of-flight (TOF) mass spectrometry had been applied to the
analysis of pyridine coal extracts [24] and of raw coals and some other extracts [25]. The
results for bituminous coal extracts are essentially the same as the results derived from
FIMS. The extracts displayed molecular mass distribution between roughly 200 and 500
amu with low intensity tailing to higher masses. The MALDI-TOF spectra of the raw
coals showed bimodal molecular mass distribution, which covered the regions 130-600
and 1300-3500 amu. Thus, the latter provides information on the mass distribution of the
macromolecular part of coal.
All the same, the question still remains whether the applied solvents that produced 3 0 -
40 wt.% extract yields were able to remove all molecular substances occurring in coal
matter. The question is still open since no data has been found on molecular mass
distribution of the high-yield extracts, obtained with the use of binary or ternary solvents.

4.3. Application o f^ N M R -fr e e induction decay measurements

In 1978, solid-state 1H NMR had been applied to observe protons in the coal by
recording their free induction decay (FID) during pulsed exposure to a high-frequency
field [26,27]. The technique was then used for observing the decay in a coal swollen with
the use of fully deuterated solvents [28,29]. The FID signal in the swollen coals had been
clearly deconvoluted into Gaussian and Lorentzian parts, corresponding to immobile
protons of the rigid macromolecules and mobile protons of rotating molecules, respec­
tively. Thus, the technique showed that coal organic matter, in the presence of solvents,
indeed consists of two entirely different parts with regard of their rotational freedom. The
calculations indicated approximately 40 wt.% content of mobile molecules in the coal
swollen by pyridine-ds [28]; the value that was close to the extract yield obtained with the
use of ethylenediamine/pyridine mixture [17]. In general, the experiments showed the
following: when the same solvent was used for extraction and in 1H NMR experiments,
the content of mobile molecules indicated by the latter was always higher compared with
the extract yield [28]. The discrepancy was ascribed to the occurrence of some molecules
in pores having narrow orifices. The solvents could pass the orifices and make “mobile”
the inside coal molecules; however, the molecules could not be extracted due to their large
size and too narrow passages [30].
The described interpretation of the NMR spectra was the subject of the debate in print
organized by Peter Given [31]. One of the conclusions was that the hydrogen atoms of
some alkyl substituents of the network might also show high “mobility” comparable with
the mobile protons of the relatively small molecules.
30 A. Marzec / Fuel Processing Technology 7 7 -78 (2002) 25-32

The results of extraction experiments, FI mass spectrometry and 1H NMR studies were
then used by Marzec and co-worker [30,32] for working out the mechanism of coal
extraction and swelling as well as the model of macromolecular/molecular structure of
coal [32]. The model was tested and confirmed by others [33-36] and likewise contra­
dicted [37]. With regard to swelling, it should be pointed out that Suuberg et al. [38] had
presented in 1994 an advanced thermodynamic concept of coal swelling.

5. Nature of attractive forces in coal organic matter

The conclusion from the extraction studies [16,20] was that the molecular substances
are attached to the macromolecular network by means of forces between electron-donor
and electron-acceptor centers occurring in molecular as well as macromolecular parts of
coal organic matter.
In late 1990s, Larsen and Gurewich [39] as well as Sakurovs [40] pointed out that
whole coal organic matter is most likely held together by all of the attractive forces
experienced by organic compounds. In his statement, Sakurovs included electrostatic
interactions (ionic and hydrogen bonds) and electrodynamic interactions [40]. The latter
occur when there are instantaneous variations in the distribution of electric charge on the
species. The variations generate net attractive forces such as the van der Waals and P i-P i
interactions. The presence in coal of various oxygen groups that may form hydrogen
bonds, which in general have rather high energy ( ~ 20 kJ mol"1), support the view that
the H bonds may be major attractive forces in coal organic matter [39]. On the other hand,
aromatic rings are most abundant structures in bituminous coals and their P i-P i
interactions may commonly occur. It is hard to assess the energy of P i-P i interactions
in coal since it strongly depends on mutual position of the aromatic rings (face-to-face,
displaced, etc.) and on the size of aromatic units—the conclusion was derived from
molecular and quantum mechanics calculations for model aromatic hydrocarbons [41].
Later studies were aimed at experimental discriminations of the nature of attractive
forces occurring in coal organic matter. Larsen [39] proposed a method of counting the
hydrogen bonds in coals on the basis of swelling measurements. For swelling, he used
series of mixtures of increasing content of pyridine in a nonpolar solvent; pyridine was
assumed to destroy all hydrogen bonds occurring in coal. In 1998, an improved IR method
of observation of the type of hydrogen bonds in coal by FTIR was proposed [42]. The
method includes the removal of water from coal in a high vacuum chamber with CaF2
windows followed by in situ scanning of the IR spectra. Five types of hydrogen bonds
were found in coal and its extracts. The thermal stability of the bonds was investigated by
means of in situ pyrolysis FTIR. The stability follows the order: O H -ether 0 > self-
associated OH and cyclic OH>OH—N >O H -Pi electrons.
With the aim to investigate the electrodynamic forces in coal, Sakurovs [40] applied
proton NMR technique. He observed an increase of coal mobility in the presence of a
number of aromatic hydrocarbon additives. The results were interpreted in terms of
reduced interactions in coal that are electrodynamic in nature.
Molecular dynamics calculations had been used for simulating interactions between
coal and solvents [43]. The results o f the simulation indicated that “ cooperative
 A. Marzec / Fuel Processing Technology 7 7 -78 (2002) 25-32 31

interactions through hydrogen bonds and aromatic-aromatic interactions at several sites of

6. Conclusions

Available data (MALDI-TOF) for a few coals indicate that the mass distribution of the
macromolecular part of coal is in the 1300-3500 amu range. However, the structure of the
macromolecular part still poses an enigma.
The question whether at least some macromolecules form three-dimensional covalently
linked network, the sort that cannot be dissolved in any organic solvent, is still open. Some
binary or ternary solvent systems recently applied can dissolve as high as 70 wt.% of
bituminous coal organic matter. How can we know that even more efficient solvent
mixtures would not be found? On the other hand, can we be sure that strong basicity of
highly efficient solvents does not decompose some covalent bonds? Moreover, solubility
of coal organic matter does not exclude the possibility of the occurrence of other types of
three-dimensional polymeric structures in the material. For example, some hyperbranched
polymers, which form indeed, three-dimensional structures, are easily soluble in common
organic solvents [44]. The same refers to some entangled structures.
There is no question that significant amount of various compounds of the — 100 -500
amu range occur in bituminous coal matter. However, their total content is not known in
any coal.
Nucleophilic property is a major factor that determines solvent effectiveness in coal
extraction. A major contributor to the mechanism of coal extraction is a substitution
phenomenon in which electron-donating solvent replaces the coal electron-donor partic­
ipant in the intermolecular forces occurring in coal. Other factors that influence the solvent
effectiveness is shape and size of solvent molecules, parameters which determine the
solvent ability to penetrate the most inner spots of coal organic matter.
Most likely, all the types of intermolecular forces hold together coal organic matter. As
yet, however, only hydrogen bonds can be directly detected.

Acknowledgements

The title of the paper “Towards an Understanding of the Coal Structure” has been
taken from “An Essay on the Organic Geochemistry of Coal” by P.H. Given [6], The title
accurately describes the past as well as the present situation in coal structure studies.

References

[1] R.M. Davidson, Molecular Structure of Coal, IEA Coal Research, London, 1980.
[2] D.W. van Krevelen, Fuel 61 (1982) 786-790.
[3] M.W. Haenel, Fuel 71 (1992) 1211-1223.
[4] K.H. van Heek, Fuel 79 (2000) 1-26.
32 A. Marzec / Fuel Processing Technology 7 7 -78 (2002) 25-32

[5] D.W. van Krevelen, Coal: Typology— Physics— Chemistry— Constitution, 3rd edn., Elsevier, Amsterdam,
1993, pp. 777-810.
[6] (a) P.H. Given, in: M.L. Gorbarty, J.W. Larsen, I. Wender (Eds.), Coal Science, vol. 3, Academic Press,
London, 1984, p. 179.
(b) P.H. Given, in: M.L. Gorbarty, J.W. Larsen, I. Wender (Eds.), Coal Science, vol. 3, Academic Press,
London, 1984, pp. 179-190.
(c) P.H. Given, in: M.L. Gorbarty, J.W. Larsen, I. Wender (Eds.), Coal Science, vol. 3, Academic Press,
London, 1984, pp. 190-192.
[7] A. Oberlin, et al., Carbon 18 (1980) 347.
[8] J.N. Rou/aud, A. Oberlin, in: H. Charcosset (Ed.),Advanced Methodologies in Coal Characterization,
Elsevier, Amsterdam, 1990, pp. 311-351.
[9] J.W. Larsen, J. Kovac, in: J.W. Larsen (Ed.), Organic Chemistry of Coal—ACS Symp., vol. 71, Am. Chem.
Soc., Washington, DC, 1978, pp. 36-49.
[10] L.M. Lucht, N.A. Peppas, in: B.R. Cooper, L. Petrakis (Eds.), Chemistry and Physics of Coal Utilization,
Am. Inst. Phys., New York, 1980, pp. 28-48.
[11] M. Krzesinska, Energy Fuels 15 (2001) 930-935.
[12] M. lino, T. Takanohashi, H. Ohsuga, K. Toda, Fuel 67 (1988)1639-1647.
[13] G. Dyrkacz, Energy Fuels 15 (2001) 918-929.
[14] G.R. Dyrkacz, C.A.A. Bloomquist, Energy Fuels 15 (2001) 1409-1413.
[15] M. lino, Fuel Process. Technol. 62 (2000) 89-101.
[16] A. Marzec, M. Jozwa, K. Betlej, M. Sobkowiak, Fuel Process. Technol. 2 (1979) 35-44.
[17] D. Bodzek, A. Marzec, Fuel 60 (1981) 47-51.
[18] A. Marzec, Proc. Intern. Conference on Coal Science, Gluckauf, Essen (1981) 98-103.
[19] V. Gutmann, The Donor-Acceptor Approach to Molecular Interactions, Plenum, New York, 1978.
[20] B. Chawla, B.H. Davis, Fuel Process. Technol. 23 (1989) 133-148.
[21] A. Marzec, H.R. Schulten, Fuel 73 (1994) 1294-1305, and references therein.
[22] N. Simmleit, H.R. Schulten, Y. Yun, H.L.C. Meuzelaar, in: H.L.C. Meuzeelar (Ed.), Advances in Coal
Spectroscopy, Plenum, New York, 1992, pp. 295-338.
[23] J. Pajak, A. Marzec, D. Severin, Fuel 64 (1985) 64-67.
[24] J.E. Hunt, R.E. Winans, Prepr. - Am. Chem. Soc., Div. Fuel Chem. 40 (3) (1995) 449-452.
[25] K. Miura, M. Shimada, K. Mae, H.Y. Sock, Fuel 80 (2001) 1573-1582.
[26] T. Yokono, Y. Sanada, Fuel 57 (1978) 334-341.
[27] L.J. Lynch, D.S. Webster, Fuel 58 (1979) 235-245.
[28] A. Jurkiewicz, A. Marzec, N. Pislewski, Fuel 61 (1982) 647-650.
[29] A. Marzec, A. Jurkiewicz, N. Pislewski, Fuel 62 (1983) 996-998.
[30] A. Marzec, W. Kisielów, Fuel 62 (1983) 977-978.
[31] P.H. Given, A. Marzec, W.A. Barton, L.J. Lynch, B.C. Gerstein, Fuel 65 (1986) 155-163.
[32] A. Marzec, Fuel Process. Technol. 14 (1986) 39-46.
[33] B. Kamieński, M. Pruski, B.C. Gerstein, P.H. Given, Energy Fuels 1 (1987) 45-50.
[34] K.E. Singleton, R.G. Cooks, K.V. Wood, A. Rabinovich, P.H. Given, Fuel 66 (1987) 74-82.
[35] M. Monthioux, P. Landais, Fuel 66 (1987) 1703-1708.
[36] L.J. Lynch, R. Sakurovs, D.S. Webster, P.J. Redlich, Fuel 67 (1988) 1036-1041.
[37] M. Nishioka, M.L. Gorbaty, Energy Fuels 4 (1990) 70-73.
[38] E.M. Suuberg, Y. Otake, M.J. Langner, K.T. Leung, I. Milosavljevic, Energy Fuels 8 (1994) 1247-1262.
[39] J.W. Larsen, I. Gurewich, Energy Fuels 10 (1996) 1269-1272, and references therein.
[40] R. Sakurovs, Energy Fuels 12 (1998) 631-636.
[41] A. Marzec, Carbon 38 (2000) 1863-1871.
[42] C. Chen, J. Gao, Y. Yan, Energy Fuels 12 (1998) 446-449.
[43] T. Takanohashi, M. lino, K. Nakamura, Energy Fuels 12 (1998) 1168-1173.
[44] A. Hult, M. Johansson, E. Malmstrom, in: Advances in Polymer Science, vol. 143, Springer V., Berlin, 1999,
pp. 1-33.