Fuel Processing Technology 85 (2004) 1671 1687

www.elsevier.com/locate/fuproc

Analysis of solvent extracts from coal liquefaction in

a flowing solvent reactor
Wen-Ying Li a,*, Jie Feng a, Ke-Chang Xie a, R. Kandiyoti b
a

Key Laboratory of Coal Science and Technology, Taiyuan University of Technology,

Ministry of Education and Shanxi Province, No. 79 Yingze West Street, Taiyuan 030024, PR China
b
Department of Chemical Engineering and Chemical Technology, Imperial College, University of London,
London SW7 2BY, UK
Accepted 1 December 2003

Abstract
Point of Ayr coal has been extracted using three solvents, tetralin, quinoline and 1-methyl-2pyrrolidinone (NMP) at two temperatures 350 and 450 jC, corresponding approximately to before
and after the onset of massive covalent bond scission by pyrolysis. The three solvents differ in
solvent power and the ability to donate hydrogen atoms to stabilise free radicals produced by
pyrolysis of the coal. The extracts were prepared in a flowing solvent reactor to minimise secondary
thermal degradation of the primary extracts. Analysis of the pentane-insoluble fractions of the
extracts was achieved by size exclusion chromatography, UV-fluorescence spectroscopy in NMP
solvent and probe mass. With increasing extraction temperature, the ratio of the amount having big
molecular weight to that having small molecular weight in tetralin extracts was increased; the tetralin
extract yield increased from 12.8% to 75.9%; in quinoline, increasing extraction temperature didnt
have an effect on the molecular weight of products but there was a big increase in extract yield. The
extracts in NMP showed the enhanced solvent extraction power at both temperatures, with a shift in
the ratio of larger molecules to smaller molecules with increasing extraction temperature and with the
highest conversion of Point of Ayr coal among these three solvents at both temperatures. Solvent
adducts were detected in the tetralin and quinoline extracts by probe mass spectrometry; solvent
products were formed from NMP at both temperatures.
D 2004 Elsevier B.V. All rights reserved.
Keywords: Coal liquefaction; Size exclusion chromatography; UV-F; Probe-MS

* Corresponding author. Tel./fax: +86-351-601-8453.

E-mail address: ying@tyut.edu.cn (W.-Y. Li).
0378-3820/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.fuproc.2003.12.012

1672

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

1. Introduction
Similarities between the early stages of thermal breakdown during coal pyrolysis and
liquefaction have been discussed in considerable detail [1]. Electron spin resonance based
studies suggest massive covalent bond scission to begin taking place at temperatures
between 350 and 375 jC (occasionally nearing 400 jC)depending on the nature of the
coal sample and the heating rate. At these temperatures, the concentration of both reactive
and relatively stable free radicals tend to increase sharply [2], with the sharp increase in
stable free radical concentrations arguably representing the debris of completed covalent
bond scission reactions.
It is at these temperatures that predominant reaction pathways in pyrolysis and
liquefaction are thought to diverge, depending on local reaction conditions (availability
of local hydrogen). The eventual outcome of the process appears to be related to
competition between relative rates of two (probably overlapping) classes of reactions:
(i) free radical quenching, resulting in product release, as extract or tar plus gaseous
products, and (ii) recombination reactions leading to resolidification within the coal mass.
The enhanced presence of locally available hydrogen would be expected to shift the
competition between these two classes of reactions in favour of greater product release and
slower resolidification processespresumably by quenching free radicals. The analogous
roles were played by locally evolved hydrogen during the pyrolysis of liptinites and the
presence of excess H-donor solvent during the liquefaction of coals has already been
discussed [1]. In both cases, the presence of excess (and in situ available) hydrogen
appears to delay resolidification.
As an aside, variations in the relative rates of these competing processes are viewed as
giving rise to changes in product distributions with changing heating rates [1,3]. In the
presence of excess hydrogen, the sensitivity of the product distributions to changes in
heating rate are found to be much reduced (cf. also Ref. [4]).
In a set of experiments closely related to these arguments, we conducted liquefaction
experiments in a tubular reactor, where solvent was continuously forced through a fixedbed of coal, removing released extract molecules into a cold zone for rapid quenching [5].
By segregating extract molecules from parent coal particles, this reactor configuration
allows detailed characterization of freshly extracted coal-derived molecules [6]. Extracts
prepared in this reactor gave the largest molecular mass (MM) distribution compared to
those obtained from extracts prepared in a mini-bomb reactor and the Point of Ayr pilot
plant.
The characterization of the extracts has previously been carried out at the time of the
original experiments by size exclusion chromatography (SEC) using tetrahydrofuran
(THF) as the eluent. However, the use of THF in SEC has been shown to promote
adsorption of coal-derived materials on the column packing, giving results not related to
size exclusion, as well as being a poor solvent compared to 1-methyl-2-pyrrolidinone
(NMP) [1,7]. Direct comparisons of chromatograms obtained by SEC of the same samples
in NMP and in THF are presented in Ref. [8].
NMP was first proposed by Lafleur and Wornat [9] as eluent for SEC; it is more polar
and appears to completely dissolve materials as heavy as coal tar pitches. Since adopting
NMP as eluent for SEC, good qualitative agreement was obtained between Matrix

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

1673

Assisted Laser Desorption Ionisation-Mass Spectroscopy (MALDI-MS) and SEC where

NMP was used as eluent [10]. Work in this laboratory involving the tandem use of SEC
and MALDI-MS has given rise to the identification of very large ranges of MM in coal
extracts and pitch samples [11].
The object of this work was to repeat the flowing solvent reactor (FSR) extractions
using tetralin, quinoline and NMP at two temperatures 350 and 450 jC (respectively
before and after the onset of massive pyrolytic covalent bond cleavage) with characterization of products by the newer methods based on NMP solutions and also to try to
distinguish the effect of solvent power and H-donor ability on coal liquefaction extracts by
using NMP as one of the extraction solvents.

2. Experimental
2.1. Sample
Point of Ayr (PoA, UK) coal sample (% w/w daf basis: C: 84.5, H: 5.4, N: 1.8, S: 1.5,
O: 6.1 (by diff.), ash 10.6) was ground and sieved to separate the 106 150 Am fraction in
a nitrogen-filled glove box. Samples were dried in a vacuum oven at 35 jC for 16 h and
stored under nitrogen in a freezer until required.
2.2. Preparation of PoA coal liquefaction extracts
The FSR system has been described in detail elsewhere [5]. The reactor consists of a
direct electrically heated tubular vessel; a fixed bed of about 150 mg coal mixed with acidwashed sand (106 150 Am particle size) is plugged between two matted layers of wire
mesh, positioned in the upper part of the reactor. The steel powder (Fe/Cr18/Ni10/Mo3) is
used when NMP is served as solvent. The continuous flow of solvent was preheated in the
lower section of the reactor to the temperature of the upper section and forced through the
sample-bed. Solvent flow was maintained at 0.9 ml/s at a pressure 7 MPa. In the present
study tetralin, quinoline and NMP have been used as solvent. Samples were heated to 350
or 450 jC at 5 K s 1 and kept at the peak temperature for a holding period of 400 s.
After an experiment, residual extract and solvent in the reactor were washed through
with a mixture of chloroform and methanol (4:1 ratio). To remove traces of solvent, the
reactors are heated in a furnace to 250 jC in the presence of a nitrogen purge for at least 16
h. The reactors were then cooled, weighed and the sample weight loss calculated as the
change in weight of the reactor (plus contents) as a percentage of the weight of coal used.
Possible sources of error in this measurement have been evaluated [5]; better than 1%
repeatability is achieved in these experiments.
The products of reaction include solvent-derived material [12] as well as coal extract.
Coal extracts were recovered by vacuum-rotary evaporation. Small-MM products usually
distil off along with the solvent but higher boiling point solvent-derived dimers and related
adducts remain with the product. However, these have been shown to be pentane-soluble;
separation of the pentane-insoluble (PI) fraction was found to concentrate the greater part of
the coal extract in the PI fraction with minimal interference from solvent-derived materials.

1674

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

The PI fractions were prepared by adding the extract solution to pentane (minimum
volume ratio 1:20), removing the pentane (plus pentane solubles) layer and drying the
extract. The PI fraction of the extracts was left to dry in a vacuum oven at 80 jC, to
remove the traces of tetralin or quinoline, as identified to remain in the extract eventually
by GC. But for coal extracts in NMP solvent, there are no PI fractions because of NMP
solvent property. NMP extracts could be obtained directly by drying and removing NMP
solvent (mostly) in vacuum oven at 120 jC.
2.3. Size exclusion chromatography (SEC)
A Polymer labs Mixed-D (5 Am) column was used at 80 jC with Perkin-Elmer isocratic
pump LC250. Three detectors were connected to the column, a UV Perkin-Elmer LC 290 at
450 nm, an Applied Biosystems diode array detector at 280, 300, 350 and 370 nm and a
Polymer Lab Evaporative Light Scattering detector (ELS) were employed to characterize the
fractions. The eluent NMP was pumped at the flow rate of 0.5 ml/min with the pressure 11
14 MPa. Data from the ELS detector were collected using a Perkin-Elmer 1020 data system
and overlaid using Microsoft Excel software. The data from UV detector were acquired in
new software, which allows the analysis of the sample simultaneously at five different
wavelengths namely 280, 300, 350, 370 and 450 nm by the means of a single injection of the
sample in the SEC column. The UV detectors respond only to aromatic material, and the
NMP solvent is not considered a good solvent for aliphatics, particularly alkanes. The ELS is
considered to give a more quantitative response to the material excluded for the column
porosity than the UV detectors, and would respond to loose aliphatic material.
The calibration curve against polystyrene standards for the Mixed-D column used in
this work has been presented elsewhere [13]. In terms of these standards, the nominal
linear range of the calibration curve (log of MM vs. elution time) was found to extend up
to between 58 and 200,000 u; larger molecular mass (MM) polystyrene standards, up to
1,850,000 u, were observed to elute from this column at shorter times, with a different
linear relation with time.
Evidence presented elsewhere [11,14] has shown that when NMP is used as eluent, the
calibration of the column against polystyrene standards serves as a good indicator of MM
for coal-derived materialsand is relatively independent of molecular structure, at least
for the material retained by the column. Direct confirmations of MM determined by this
method have recently been obtained, in the form of an absolute calibration against
MALDI-MS readings on the same samples.
2.4. UV-fluorescence spectroscopy
A Perkin-Elmer LS50 Luminescence spectrometer was used in the static-cell mode with
NMP as solvent; a quartz cell with 1 cm path-length was used. NMP has very little
fluorescence but is opaque at 254 nm, becoming partially transparent by 260 270 nm.
Procedures for acquiring UV-fluorescence spectra have been described elsewhere [15].
Synchronous spectra were acquired at a scan speed of 240 nm min 1 and a constant
wavelength difference of 20 nm. Emission and excitation spectra were also obtained but
have not been shown.

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

1675

2.5. Probe mass spectroscopy

The pentane insoluble (PI) fraction of the extracts were also characterized by Probe MS
using a Joel JMS-AX505W double-focusing mass spectrometer equipped with a standard
solid injection probe and electron-impact ionisation. Sample was inserted into the glass
sample tip and heated in the ion source vacuum to 400 jC at a heating rate of 64 jC/s
while more spectra were acquired over mass range m/z 30 800.

3. Results and discussions

3.1. PoA coal liquefaction yield
Table 1 presents PoA coal liquefaction yield from liquefaction experiment, using PoA
coal at 350 and 450 jC in tetralin, quinoline and NMP. All data have been corrected by the
weight loss of sand in the blank run. A different sample of Point of Ayr coal has been used
in these experiments and exact agreement with previous results [5] could not be expected,
although the trends were roughly consistent.
At 350 jC, sample weight loss data in quinoline were larger than those in tetralin. As
discussed above, at up to 350 jC the process is mainly extraction, with little (if any)
covalent bond scission taking place. Tetralin as well as quinoline would, under these
conditions, merely act as solvents, and tetralin would not have played a significant Hdonor role. The greater weight loss at this temperature would, therefore, indicate the
somewhat greater solvent power of quinoline, compared to tetralin, for coal-derived
materials.
At 450 jC, the different trends as previously observed with slightly greater weight loss
in quinoline than in tetralin become apparent, although tetralin reflected its H-donor ability
at higher temperature; as above it seems possible to invoke hydrogen induced bond
scission, the differences were big. With increase of reaction temperature, the solvent power
is increased, which results in the increase of extract yields in all these three solvents. The
range of the weight loss among three solvents is from 75.9% (tetralin) to 83.6% (NMP).
The weight loss of coal in NMP at both 350 and 450 XC was considerably higher than
the weight losses in tetralin and quinoline. Clearly, NMP is a good solvent for all but the
most aliphatic of coal-derived materials. Thus potentially soluble free radicals detached

Table 1
PoA coal liquefaction yield with various solvents
Solvent

Weight loss (daf) %

Point of Ayr (with sand or steel powder)

Tetralin
Quinoline
NMP

Blank run (with sand or steel powder)

350 jC

450 jC

350 jC

450 jC

12.8
21.8
52.6

75.9
79.0
83.6

0.44
0.26
1.03

0.24
0.60
1.11

At a heating rate 5 K s

, 0.9 ml s

flow rate and 400 s holding period.

1676

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

from the main coal mass by bond scission appear to move out of the parent matrix and to
be extracted. SEC chromatograms show the presence of a fraction of relatively small
molecular material that appears to have been formed by pyrolitic cracking reactions to
form smaller molecules soluble in NMP.
3.2. SEC of PoA coal liquefaction extracts
The evaporative light scattering detector (ELS) is considered to give a more quantitative response to the material excluded for the column porosity than the UV detectors, and
would respond to loose aliphatic material. Most of the SEC chromatograms presented
below show a major peak centred at around 8 12 min representing the bulk of the largeMM material, mainly excluded from column porosity. The second peak centred around
15 20 min and the third peaks around 20 25 min representing smaller MM material
resolved by the column porosity, they are shown in Fig. 1, in which all peaks are processed
by area normalization method. Ranges of actual molecular masses corresponding to these
elution times will be discussed below. Clearly, the tetralin PI extracts have a big shift in
molecular weight from small to large with an increase of extraction temperature (Fig. 1a).
At higher temperature, quinoline solvent shows its solvent power, the relative intensity of
the quinoline PI extracts at 450 jC is much stronger than at 350 jC; and also there are
small molecules appearing in the elution time between 20 and 25 min at 450 jC, but there
isnt at 350 jC (Fig. 1b). In NMP extracts, there is an obvious difference from the other
two solvent extracts. Fig. 1c shows that there are more larger molecules formed at 350 jC
than at 450 jC. As the extraction temperature increases, the NMP extracts tend to be small
molecules. It reflects that the enhanced solvent NMP has more power to extract at low
temperature. After 25 min elution time, there is one small peak, which is supposed to be
induced by noise signal.
Fig. 2a and b present SEC chromatograms of tetralin extracts (PI) at 350 and 450 jC,
respectively, using UV adsorption detector; these chromatograms show relative intensities
of absorbance at five different wavelengths. Between the two temperatures, the order of
relative intensities of the chromatograms as a function of absorbance wavelength did not
change significantly, but the ratio of the first peak (around 8 11 min) to the second peak
(around 15 20 min) has increased, indicating that the aromatic clusters in the different
molecular sizes are significantly changed with the temperature of extraction. Polynuclear
aromatic ring systems detected at 350 and 450 nm are expected to be larger. The clusters
absorbing at these wavelengths are observed throughout the SEC profile, having more
absorbance at 450 jC. The small peak at about 22 min in Fig. 2a and b is likely to
correspond to solvent adducts retained in the pentane insoluble material [16], which are
proved to be difficult for removing totally.
In both cases, the result could be interpreted in terms of increased solubility with
increasing temperature and/or by a greater proportion of larger-MM material being
released by the coal matrix; in either case, it must be recalled that a far larger proportion
of the coal was extracted at the higher temperature. The only comparison available with
SEC in NMP from previous work is for the tetralin extract [17] obtained at 450 jC for
absorbance at 350 and 450 nm; the profiles at these two wavelengths were similar to those
given below.

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

1677

Fig. 1. SEC spectra of solvent extract using the Mixed-D column and ELS detector. (a) Tetralin extracts (PI); (b)
quinoline extracts (PI); (c) NMP extracts.

Fig. 3 presents SEC chromatograms of extracts (PI) prepared in quinoline at 350 and
450 jC, respectively. The chromatograms are qualitatively similar to those of the tetralin
extracts. At longer elution times, a second major peak was found (as in the case of tetralin
extracts), showing signal from material resolved by the column porosity, eluting between
15 and 23 min. The position of the retained peaks is similar at both temperatures, but

1678

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

Fig. 2. SEC spectra of tetralin extracts (PI) using the Mixed-D column and UV absorption detector ((1)280
nm; (2)300 nm; (3)350 nm; (4)370 nm; (5)450 nm; (a)350 jC; (b)450 jC).

shifted slightly towards smaller molecules (later elution time) compared with the tetralin
extracts. These elution times correspond to material in the range of below 400 amu
showed up only in chromatograms acquired with 350, 300 and 280 nm absorbance
wavelengths, suggesting the presence of, at most, two and/or three fused ring polynuclear
aromatic systems.
Fig. 4 presents the SEC chromatograms of extracts produced in NMP at 350 and 450
jC respectively. In both figures, the significant peaks in the excluded region (8 10 min)
are quite obvious at five different wavelengths. However, the peak from signal eluting
between 15 and 22 min was insignificant compared with those observed for extracts
produced in tetralin or quiniline. Except for the response to two short wavelengths at 280
and 300 nm, other signals are very weak; there are solvent products formed in NMP at
both temperatures. These findings strongly suggest that in NMP pyrolytic fragmentation
was the dominant mode in the formation of the products detected at either temperature.
In particular, the SEC chromatograms of coal extracts at an absorbance wavelength of
350 and 450 nm were worth further discussion. At 450 jC (Figs. 2b, 3b, 4b), the

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

1679

Fig. 3. SEC spectra of quinoline extracts (PI) using the Mixed-D column and UV absorption detector ((1)280
nm; (2)300 nm; (3)350 nm; (4)370 nm; (5)450 nm; (a)350 jC; (b)450 jC).

progressive shift of intensity from the excluded peak of the tetralin extract to the low MM
peak of the NMP extract is clear at 350 nm. The peaks around 15 20 min in tetralin and
quinoline show close correspondence and cannot be attributed solely to solvent-derived
products, as discussed above. Similarly, the pyrolysis extent at 350 jC (Figs. 2a, 3a, 4a) is
less than at 450 jC and extraction is more likely to be dependent on solvent power. The
relative intensity of the peak at 15 20 min increases as solvent power increases. Taken
together, the data show the importance of solvent power and H-donor ability in generating
an extract containing large molecules with minimum pyrolytic degradation of the extracted
material.
Briefly, at an absorbance wavelength of 450 nm, the SEC chromatograms are quite
similar to those in Figs. 2 4. An increase of reaction temperature will result in the
increase ratio of the earlier peak (large molecular mass, 8 10 min) to the later peak
(relative small molecular mass, 15 20 min); in the case of NMP, this ratio is relatively
higher at 350 jC than at 450 jC, suggesting that more solvent products formed at high
temperature decreases the NMP solvent extraction power for large molecular mass
material.

1680

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

Fig. 4. SEC spectra of NMP extracts using the Mixed-D column and UV absorption detector ((1) 280 nm; (2)
300 nm; (3) 350 nm; (4) 370 nm; (5) 450 nm; (a) 350 jC; (b) 450 jC).

3.3. UV-F analysis of PoA coal liquefaction extracts

Fig. 5 present the synchronous UV-F spectra of extracts obtained in tetralin, quinoline
and NMP at 350 and 450 jC, respectively. The difference in the UV-F profiles of the
tetralin and quinoline extract in both temperatures reflect the solvent power effects. A shift
to shorter wavelengths for quinoline reflects the increased pyrolytic fragmentation in the
absence of H-donor ability.
Synchronous UV-F spectra of the NMP extracts indicate that at 350 jC, a broader range
of aromatic clusters (300 600 nm) was detected than at 450 jC (350 600 nm). Previous
work has shown that the material of the excluded peak in SEC does not fluoresce and
therefore, the UV-F spectra apply to a small proportion of the tetralin extract, rather more
of the quinoline extract and possibly all of the NMP extract.
These findings parallel the observations from SEC discussed earlier and show the
effects of a good solvent with H-donor capability, a good solvent with no H-donor ability
and an ideal solvent lacking H-donating capability in terms of stabilising large extract
molecules.

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

1681

Fig. 5. UV-F synchronous spectra of tetralin, quinoline and NMP extracts; (a) 350 jC; (b) 450 jC.

3.4. Probe mass spectra PI from PoA coal liquefaction extracts

The peak at m/z 262 in the probe mass spectra (Fig. 6) confirms the presence of tetralin
dimers at both extraction temperatures. The MM of tetralin is 132. The intensity of this
peak however increases at elevated temperatures since tetralin tends to dimerise further at
higher temperature. At both temperatures, the presence of m/z 258 indicates the formation
of tetralin naphthalene dimers.
The formation of quinoline dimers can be seen from the probe mass spectra of the
extracts in Fig. 7. Quinoline has a mass (m/z) of 129; the signal for dimers observed
corresponds to the mass m/z 256, which shows a change in relative intensity at the two
temperatures, showing that the dimers at the higher temperature are more likely to
fragment than those formed at the lower temperature. The loss of NH3 (m/z 17) group
fragment is clear in both spectra.
From Fig. 8a,b there are no dimers or trimers formed in both temperatures. However,
more fragments formed at 450 jC than at 350 jC. The molecular mass of NMP is 99. At
lower temperature, NMP wasnt decomposed, the base peak is at m/z 99 and there is one

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

Fig. 6. Probe MS of tetralin extracts (PI): (a) 350 jC; (b) 450 jC.

1682

1683

Fig. 7. Probe MS of quinoline extracts (PI): (a) 350 jC; (b) 450 jC.

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

Fig. 8. Probe MS of NMP extracts: (a) 350 jC; (b) 450 jC.

1684

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

1685

possible molecular ion peak at m/z 129. If m/z 129 is the molecular ion peak, it should be
derived from NMP and one possible structure cleavage from m/z 129 should be -H2CO
(m/z 30). The two possible structures of m/z 129 are shown as follows:

and

With the increase of reaction temperature, the products of coal liquefaction become
complicated. Fig. 8b shows us the probe mass spectra at 450 jC, it can be found that its
structure is rather complex and destablalization (in MS ion source) compared with the
spectra at 350 jC, the base peak from m/z 99 change to m/z 44 and one H has been
removed from the NMP structure, which means lots of molecular fragments from the coal
have been connecting with NMP and lead to the destablization of this five-member circle.
One possible structure is shown as follows:

1686

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

4. Conclusions
(1) With increase of liquefaction temperature, the weight loss of coal extracted by tetralin,
quinoline and NMP is increased from 12.8% (tetralin) to 52.6% (NMP) at 350 jC, and
from 75.9% (tetralin) to 83.6% (NMP) at 450 jC. It indicates the importance of using
a solvent capable of dissolving the largely aromatic materials released from the coal at
high temperatures.
(2) SEC profiles of the extracts in tetralin at 350 and at 450 jC showed that larger
molecules were extracted at the higher temperature. For quinoline in contrast, the SEC
profile of the extract at the higher temperature indicated an increased degree of
pyrolysis and more small molecules than at the lower temperature, indicating the
importance of radical stabilisation by the H-donor solvent, tetralin in comparison with
their roughly similar solvent power. The extracts in NMP, the best solvent for
aromatics with no H-donor ability, showed an equal ratio of big and small molecules
by SEC.
(3) UV-fluorescence spectra of the extracts indicated that the tetralin and NMP extracts
were similar at the lower or higher temperature of extraction; while the quinoline
extracts showed a shift to smaller aromatic cluster sizes.
(4) Probe mass spectra of the tetralin and quinoline extract pentane-insoluble materials
indicated the presence of solvent dimers which differed in mass spectral fragmentation
pattern with extraction temperature and presumably had different structures at the two
temperatures. NMP is a good solvent at the lower temperature; with the increasing
temperature, it will react with the sample and produce some derived products.

Acknowledgements
Support for this work by the British Coal Utilisation Research Association and the
Chinese Special Funds for Major State Basic Research Projects G1999022101 is gratefully
acknowledged. W.Y. Li would like to thank the Royal Society Fellowship. The authors
would also like to thank the University of London, Kings College, ULIRS GC/MS
Spectrometry Service for Probe Mass.

References
[1] C.Z. Li, E.S. Madrali, F. Wu, B. Xu, H.-Y. Cai, A.J. Guell, R. Kandiyoti, Fuel 73 (1994) 851 865.
[2] T.G. Fowler, K.D. Bartle, R. Kandiyoti, Energy & Fuels 3 (1989) 515 522.
[3] W.S. Fong, Y.F. Khalil, W.A. Peters, J.B. Howard, Preprints of Papers Presented-American Chemical
Society. Division of Fuel Chemistry 30 (1985) 115 126.
[4] C.-Z. Li, K.D. Bartle, R. Kandiyoti, Fuel 72 (1993) 3 12.
[5] B. Xu, R. Kandiyoti, Energy & Fuels 10 (1996) 1115 1127.
[6] S.-F. Zhang, B. Xu, S.A. Moore, A.A. Herod, R. Kandiyoti, Fuel 75 (1996) 597 605.
[7] B.R. Johnson, K.D. Bartle, A.A. Herod, R. Kandiyoti, Journal of Chromatography, A 758 (1997) 65 74.
[8] S.-F. Zhang, B. Xu, A.A. Herod, R. Kandiyoti, Energy & Fuels 10 (1996) 733 742.
[9] A.L. Lafleur, M.J. Wornat, Analytical Chemistry 60 (1988) 1096 1102.

W.-Y. Li et al. / Fuel Processing Technology 85 (2004) 16711687

1687

[10] M. Domin, R. Moreea, M.J. Lazaro, A.A. Herod, R. Kandiyoti, Rapid Communications in Mass Spectrometry 11 (1997) 638 645.
[11] A.A. Herod, M.J. Lazaro, M. Domin, C.A. Islas, R. Kandiyoti, Fuel 79 (2000) 323 337.
[12] A.A. Herod, S.-F. Zhang, D.M. Carter, M. Domin, Rapid Communications in Mass Spectrometry 10 (1996)
171 177.
[13] M.J. Lazaro, C.A. Islas, A.A. Herod, R. Kandiyoti, Energy & Fuels 13 (1999) 1212 1222.
[14] C.A. Islas, PhD thesis, University of London, June, 2001.
[15] B. Xu, E.S. Madrali, F. Wu, C.-Z. Li, A.A. Herod, R. Kandiyoti, Energy & Fuels 8 (1994) 1360 1369.
[16] D. Brodzki, G. Djega-Mariadassou, C.Z. Li, R. Kandiyoti, Fuel 73 (1994) 789 794.
[17] S.F. Zhang, J.R. Shearman, M. Domin, M.J. Lazaro, A.A. Herod, R. Kandiyoti, Fuel 76 (1997) 207 214.