COAL PYROLYSIS

COAL SCIENCE AND TECHNOLOGY

Series Editor:
Larry L. Anderson
Department of Mining, Metallurgical and Fuels Engineering, University of
Utah, Salt Lake City, UT 84112, U.S.A.
Vol. 1: Geochemistry of Coal (Bouska)
Vol. 2: Fundamentals of Coal BenefiCiation and Utilization (Tsai)
Vol. 3: Coal: Typology, Chemistry, Physics and Constitution (Van Krevelen)
Vol. 4: Coal Pyrolysis (Gavalas)
COAL SCIENCE AND TECHNOLOGY 4
COAL PYROLYSIS
G.R. GAVALAS
Division of Chemistry and Chemical Engineering,
California Institute of Technology,
Pasadena, California, U.S.A.
ELSEVIER SCIENTIFIC PUBLISHING CaMPANY
Amsterdam - Oxford - New York 1982
ELSEVIER SCIENTIFIC PUBLISHING COMPANY
Molenwerf 1
P.O. Box 211,1000 AE Amsterdam, The Netherlands
Distributors for the United States and Canada:
ELSEVIER SCIENCE PUBLISHING COMPANY INC.
52, Vanderbilt Avenue
New York, N.Y. 10017
Library of Congress Cataloging in Publication Data
Gavalas, George R.
Coal pyrolysis.
(Coal science and technology
Bibliography: p.
Includes index.
1. Coal. 2. Pyrolysis. I.
TP325.G38 1982 662.6'2
ISBN 0-444-42107-
6
ISBN 044442107-6 (Vol. 4)
ISBN 044441970-5 (Series)
; 4)
Title. II. Series.
82-11374
Elsevier Scientific Publishing Company, 1982
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or
transmitted in any form or by any means, electronic, mechanical, photocopying, recording or other-
wise, w.ithout the prior written permission of the publisher, Elsevier Scientific PUblishing Company,
P.O. Box 330, 1000 AH Amsterdam, The Netherlands
Printed in The Netherlands
v
PREFACE
Prompted by the need of non-petroleum-based fuels, coal research has reemerged
to center stage after a lengthy dormant period. Pyrolysis research, in particular,
has gained considerable momentum because of its close connection to combustion,
hydropyrolysis and liquefaction. Spectroscopic and other instrumental techniques
are currently producing prodigious information about coal structure and pyrolysis
mechanisms, while modeling efforts are breaking new ground in sorting out chemical
and physical phenomena to provide a fundamental although simplified description.
The continuing generation of experimental data will lead to revisions, in
some cases drastic, of current structural and kinetic precepts. Yet, the postu-
lates and assumptions of current work provide a meaningful starting point in
elaborating theoretical descriptions of greater validity and applicability.
This monograph was written to organize recent results of pyrolysis research.
Experimental and theoretical aspects, given approximately equal weight, are dis-
cussed in the light of basic chemical and physical mechanisms. With this orienta-
tion the monograph should be useful to chemists, engineers and graduate students
with interests in coal research.
would like to express my appreciation to the copyright holders for permission
to reproduce various figures: to the American Institute of Chemical Engineers
for Figs. 4.21, 5.7, 5.14; to the Combustion Institute for Figs. 4.3, 4.14, 4.15;
to IPC Science and Technology Press, Ltd. for Figs. 4.16-4.18, 4.20, 5.1-5.6,
5.8, 5.9, 6.1, 6.6, 7.10-7.13; to Dr. W. R. Ladner for Figs. 7.6-7.9; to Dr. P. R.
Solomon for Fig. 4.1; to Mr. M. Steinberg for Figs. 7.14-7.16 and to Dr. E. M.
Suuberg for Figs. 5.10, 5.11, 6.2-6.5, 7.1-7.4.
I would also like to thank Lenore Kerner and I ~ a j d a Andlovec for typing the
manuscript and Heather Marr for drawing the figures.
Pasadena, California
April, 1982 G. R. GAVALAS
TABLE OF CONTENTS
vi i
Chapter INTRODUCTI ON
Chapter 2 CHEMICAL STRUCTURE OF COALS 3
2.1 FUNCTIONAL GROUPS 3
2.1.1 Aromatic nuclei 3
2.1.2 Aliphatic structures 4
2.1.3 Oxygen functionalities 6
2.2 STRUCTURAL AND FUNCTIONAL GROUP ANALYSIS 9
2.2.1 The Brown and Ladner method 9
2.2.2 Detailed structural analysis 10
2.2.3 Computer-assisted molecular structure construction II
2.2.4 Functional group analysis by linear programming 12
Chapter 3 THERMAL REACTIONS OF COAL 19
3.1 BOND DISSOCIATION WITH PRODUCTION OF TWO RADICALS 19
3.1.1 Activation energies 20
3.1.2 A-factors 23
3.1.3 The effect of phenolic hydroxyl groups 27
3.2 DISSOCIATION OF FREE RADICALS 28
3.3 RECOMBINATION OF ALPHA RADICALS 29
3.4 HYDROGEN ABSTRACTION 30
3.5 ADDITION OF RADICALS TO DOUBLE BONDS 31
3.6 ADDITION OF RADICALS TO AROMATIC RINGS 32
3.7 REACTIONS OF CARBOXYL AND PHENOLIC HYDROXYL GROUPS 33
3.8 CONCERTED REACTIONS 33
Chapter 4 EXPERIMENTAL TECHNIQUES AND RESULTS IN FLASH PYROLYSIS 39
4.1 EXPERIMENTAL TECHNIQUES 39
4.1.1 The entrained flow technique 39
4.1.2 The captive sample technique 41
4.1.3 The pyroprobe 44
4.2 EXPERIMENTAL RESULTS AND DISCUSSION 45
4.2.1 Weight loss 45
4.2.2 Product distribution 49
4.2.3 Char and tar composition; distribution of
sulphur and nitrogen 54
4.2.4 Effects of pretreatment and atmosphere of pyrolysis 61
4.2.5 Effect of inorganic constitutents or additives on
pyrolysis product yields 64
4.2.6 Miscellaneous techniques and results 66
4.3 PYROLYSIS PROCESSES 70
4.3.1 The COED process 70
4.3.2 The Occidental Research Corporation process 72
102
105
vii i
TABLE OF CONTEN!S (CONTINUED)
Chapter 5 HEAT AND MASS TRANSFER IN PYROLYSIS 77
5.1 PYROLYSIS AND THE PHYSICAL PROPERTIES OF COAL 7
5.1.1 rhe plastic state of coals 77
5.1.2 Changes in the porous structure of coals during pyrolysis 80
5.2 THE EFFECTS OF PRESSURE AND PARTICLE SIZE ON PRODUCT YIELDS 89
5.2.1 The effect of pressure 89
5.2.2 The effect of particle size 92
5.3 ANALYSIS OF HEAT TRANSFER 93
5.3.1 Criteria for the absence of heat transfer limitations 93
5.3.2 Analysis of combined heat transfer and kinetics 98
5.4 ANALYSIS OF MASS TRANSFER 99
5.4.1 Film mass transfer 100
5.4.2 Intraparticle mass transfer in softening coals
5.4.3 Intraparticle mass transfer in nonsoftening coals
Chapter 6 KINETIC MODELS OF COAL PYROLYSIS III
6.1 INDEPENDENT FIRST ORDER REACTIONS III
6.1.1 A single first order reaction 112
6.1.2 Several first order reactions for weight loss 113
6.1.3 Several first order reactions for individual products 115
6.2 COMPETING REACTIONS 122
6.2.1 Model of Unger and Suuberg 123
6.2.2 Model of Solomon 125
6.3 DETAILED CHEMICAL MODELS 128
6.3.1 A detailed kinetic model 129
6.3.2 Further ideas on kinetic modeling 134
Chapter 7
7.1
7.2
7.3
7.4
7.5
7.6
HYDROPYROLYSIS 138
CAPTIVE SAMPLE EXPERIMENTS 138
PACKED BED EXPERIMENTS 143
MODIFIED CAPTIVE SAMPLE EXPERIMENTS
ENTRAINED FLOW EXPERIMENTS 150
MODEL COMPOUND STUDIES 154
MODEll NG 155
146
REFERENCES 159
AUTHOR INDEX 164
SUBJECT INDEX 167
Chapter 1
INTRODUCTION
During the last decade, developments of both experimental and theoretical nature
have advanced our knowledge of coal pyrolysis to a degree that justifies a syste-
matic exposition on the subject. Noteworthy among experimental advances and refine-
ments have been the design of equipment for rapid but controlled heating and the
analysis and characterization of pyrolysis products by chromatographic and spectro-
scopic measurements. Modeling advances worth mentioning include the estimation
of rate parameters by group additivity techniques, improved capabilities for
numerical handling of complex reaction systems and improved understanding of the
coupling between mass transfer and chemical reactions.
Despite the aforementioned advances, the bulk of available information is of
experimental and descriptive nature while the understanding of fundamental mechan-
isms remains largely qualitative. In surveying and summarizing experimental in-
formation, we have been selective rather than exhaustive, emphasizing studies
involving broad and controlled variation of experimental conditions. The dis-
cussion of experimental results and kinetic models has concentrated on fundamental
issues such as reaction mechanisms and interaction between reactions and mass
transfer. This choice of subject matter clearly reflects the personal interests
of the author.
Early pyrolysis research was concentrated on coal carbonization, i.e. slow heat-
ing of dense coal samples, conditions pertaining to coke making. By contrast,
recent work has emphasized rapid heating of dilute coal samples which has been
amply termed "flash pyrolysis". Behind this shift in emphasis have been the de-
sire to elucidate the role of pyrolysis in combustion and gasification and the
possibility of applying pyrolysis to convert coal to liquid fuels or chemicals.
The material in this monograph is mainly addressed to pyrolysis under conditions
of rapid heating.
Chapters 2 and 3 are devoted to the chemistry of pyrolysis. In Chapter 2,
recent spectroscopic and wet chemical information is utilized to describe coal's
structure in terms of functional groups. Since most of the experimental data
concern coal derived liquids, extrapolation to solid coal involves considerable
uncertainty. However tentative it may be, this extrapolation is necessary for
mechanistic analysis of pyrolysis. Having established or postulated the pertinent
functional groups, it is possible to identify with considerable confidence the
important elementary thermal reactions. Basic organic chemistry and kinetic
studies with model compounds, provide a firm basis for this purpose. Chapter 3
contains a survey of the energetically favored elementary reactions and discusses
techniques for estimating their rate parameters. The estimation techniques are
2
illustrated with numerous examples. However, a number of elementary reactions
defy estimation by these established predictive techniques. This is the case, for
example, with reactions which are limited by diffusion in the condensed phase.
The bulk of experimental results are presented in Chapter 4. The data consist
of the yield of various pyrolysis products as a function of temperature and
pyrolysis time under conditions that minimize the influence of heat and mass
transfer limitations. A brief discussion of commercial process development efforts
is also included in this chapter. Chapter 5 is devoted specifically to heat dnd
mass transfer limitations as manifested by the dependence of product yields on
pressure and particle size. Although :he chapter cuntilns experimental and modeling
results, the modeling pat't is limited to delineating conditions under which trans-
port phenomena become important.
In the presentation of experimental results in chapters 4 and 5, a distinction
is made between softening coals, primarily high volatile bituminous, and nons oft-
ening coals e.g. subbituminous or lignites. This rudimentary distinction does not
do justice to the v a r i e ~ of composition and properties within each of the two
categories. However, it is a point of departure and has particular significance
relative to the evolution of porous structure and plastic properties both of
which strongly influence mass transfer rates.
Chapter 6 contains a survey of kinetic models of pyrolysis with emphasis on
underlying assumptions rather than mathematical details. A brief discussion is
also given about the ability of each model to describe experimental data. This
ability does not by itself imply the validity of the underlying assumptions.
Product yields are smooth and gently sloping functions of operating conditions,
that could be fitted by many phenomenological models with reasonable kinetic
form, provided the rate parameters are in each case suitably adjusted. Quite
often the scatter in the data masks subtle distinctions between alternative
models. Phenomenological models have reached a maturity that allows their
application to coal combustion or gasification. By contrast, the one or two
detailed chemical models discussed in this chapter are still at a stage of flux
and are not ripe for practical applications.
The final Chapter, 7, contains a brief survey of experimental data and model-
ing work on hydropyrolysis, the heating of coal in a hydrogen atmosphere. This
variant of pyrolysis is receiving increased attention as one of the most promis-
ing routes to chemicals from coal.
3
Chapter 2
CHEMICAL STRUCTURE OF COALS
2.1 FUNCTIONAL GROUPS
The scope of any systematic description of coal structure is seriously limited
by the large variety of coals and the variability in the chemical properties of
different macerals in a s ; n g ~ e coal. Nevertheless, such a description becomes
meaningful if it is posi-lJlated that the chemical properties of coal are determined
by a modest number of functional groups common to all coals. Coals are then assumed
to differ by their content of these functional groups. Although different macerals
(vitrinite, fusinite, exinite, etc.) differ markedly in composition and reactivity,
in this treatise we will not take into account such differences, since most of
the pyrolysis data till now have been obtained with whole coals. However, in as
much as vitrinite is the dominant maceral (more than 70% in most coals), the
structural features discussed below essentially refer to the vitrinite component.
The idea of representing different coals by a common set of functional groups has
not been fully justified but is generally compatible with the experimental evidence
discussed in the next section.
The reactivity of coal in pyrolysis, hydropyrolysis and liquefaction can be
characterized by six classes of functional groups: aromatic nuclei, hydroaromatic
structures, alkyl chains, alkyl bridges and oxygen groups. Sulfur and nitrogen
groups are important from the standpoint of pollutant formation but do not play
a significant role in the thermal reaction mechanisms, if only due to their gen-
erally low content. Each of these classes of functional groups will be discussed
below drawing mainly from the recent literature.
2.1.1 Aromatic nuclei
Nuclear magnetic resonance spectroscopy (lH and 13C) of coals and coal liquids
has established that 40-75% of carbon in subbituminous and bituminous coals is
aromatic, the aromaticity increasing with rank. The ability to determine the
aromaticity reasonably accurately has blunted or made irrelevant earlier arguments
about the nature of the carbon skeleton in coal. Aromatic rings of various sizes
are important building blocks of coal structure and maintain their integrity when
coal is heated at temperatures as high as 700C. Even when the aromatic content
is as low as 40%, the reactivity of the aliphatic carbon is strongly influenced
by proximal aromatic nuclei.
A variety of techniques have been employed to determine the size of the aro-
matic nuclei. Early studies employing X-ray diffraction spectroscopy, reviewed
by Speight and Given (refs. 1,2) suggested that coals with up to 87% carbon (this
4
includes lignites, subbituminous and bituminous coals) contain mainly units with
3 to 5 condensed rings and a smaller but appreciable amount of single and double
rings. This conclusion is subject to considerable uncertainty in view of a num-
ber of assumptions that had to be made concerning the geometric arrangement and
the disposition of aliphatic substituents. Fluorescence spectroscopy of bituminous
coal extracts (ref. 3) and comparisons with model compounds suggest that the
more abundant aromatic nuclei consist of three or more condensed rings. P o l a r o ~
graphic analysis of SRC liquids of a high volatile C bituminous coal has indicated
a small content of units with more than two condensed rings (ref. 4). Ruberto,
et al. (ref. 5) applied a variety of techniques including lH nmr, U.V. and mass
spectroscopy to a solvent extracted subbituminous coal to conclude that the aro-
matic units consist of two and three condensed rings. Since these analyses were
performed on coal liquids, drawing inferences about the structure of solid coal
involves some uncertainty.
The size of the aromatic nuclei can also be estimated from the elemental compos-
ition and lH nmr spectra of coal extracts or other coal derived liquids by a pro-
cedure known as "structural analysis", described in the next section. In as much
as structural analysis employs several assumptions, these estimates are subject
to additional uncertainty. Tingey and Morrey (ref. 6) have estimated an average
size of 5-8 condensed rings for the extracts of a subbituminous and two bituminous
coals. vJhitehurst (ref. 4) has investigated the asphaltol fraction of a liquefied
bituminous coal and proposed structures consisting of one, two and three condensed
rings. Gavalas and Oka (ref. 7) have applied the Brown and Ladner structural
analysis (see next section) to the extract of a bituminous coal to estimate nucleus
size of two to three condensed rings.
The consensus emerging from the aforementioned studies is that lignites and
subbituminous coals contain nuclei with one to three condensed rings, while bi-
tuminous coals contain mostly two to four condensed rings, the degree of conden-
sation increasing with the coal rank.
Two and three condensed ring nuclei found in substituted form in coal derived
liquids include naphthalene, indene, biphenyl, furan, pyrrole, pyridine, benzo-
thiophene, phenanthrene, anthracene and various three-ring heterocycles of oxygen,
sulfur and nitrogen. Some of the rings may be partly hydrogenated constituting
what are known as "hydroaromatic" structures.
2.1.2 Aliphatic structures
The aliphatic segments of coal include aliphatic chains (methyl, ethyl, etc.),
bridges (methylene, ethylene) and hydroaromatic structures. The information
available about aliphatic structures derives mainly from proton and 13C magnetic
resonance spectra and from selective oxidation, especially by the recently de-
veloped technique of Deno and coworkers. (refs. 12-14).
5
Proton and 13C magnetic resonance has been fruitfully applied to coal derived
liquids: pyrolyzates, extracts, and hydrogenation products that are all modified
fragments of the original coal. The modification is relatively small in the ex-
traction, or H-donor mediated extraction at low conversions, somewhat larger in
flash pyrolysis, and rather severe in liquid phase hydrogenation to distillate
products.
Figure 2.1. below illustrates the information derived from nuclear magnetic
resonance. The lH nmr spectrum distinguishes three major types of hydrogen:
I' 2'
r:Q:rg(CH
2
- CH"OO
3
OO:
3
3' 4' 5'
CH
2
- CH
2
- CH
3
6
C
6
6'
Fig. 2.1. Structural information derived from nmr spectra.
H
ar
+ 0' aromatic and phenolic hydrogen, H
a
, hydrogen at position a as 1, 2 , 1 ~ ,
2 ~ , 3 , 3 ~ , 6 , 6 ~ of Figure 2.1 and H
S
+' hydrogen at position S and further from the
ring as 4 , 4 ~ , 5 ~ of Fig. 2.1. In recent years it has been possible to further re-
solve the alpha and beta region. For example, Bartle et al,(refs. 8,9) distin-
guished two types of a and several types of S hydrogen in the lH nmr spectra of
?upercritical coal extracts. Similar distinctions were made in the spectra of
coal hydrogenation liquids by Yokohama et al. (ref. 10). Although high resolution
lH nmr spectra provides very detailed information on hydrogen types, the inter-
pretation of the spectra is not without some ambiguity. For example, the dis-
tinction between the hydrogens 1~ and 3 in Fig. 2.1 is somewhat uncertain.
13C nmr spectra have been similarly used to distinguish among Car' and several
types of C
ex
and C
13
(refs, 9,11), however, drawing quantitative results from these
G
spectra involves a larger error than in the case of lH spectra.
Deno and coworkers (refs. 12-14) demonstrated with model compound studies
summarized in Table 2.1 that an aqueous solution of H
2
0
2
and trifluoroacetic acid
selectively oxidizes the aromatic part of a molecule, leaving much of the ali-
phatic part intact. Amonq the model compounds listed in Table 2.1 only Dgives
tri- and tetra-carboxylic acids. Thus, the presence of the latter acids in the
oxidation proDucts of the Illinois #6 coal was regarded as evidence that D or
similar structures constitute a large fraction of the structural units of this
coal. Unfortunately, no work was performed with compounds like
to assess the possible role of ethylene bridges. The distinction between methyl-
ene and ethylene bridges is important from the standpoint of reactivity as will
be discussed in a subsequent chapter.
As shown in Table 2.1, the oxidation products from the subbituminous Wyodak
coal are the same as from the Illinois #6 but the relative amounts are very dif-
ferent. The Wyodak coal gave a lower yield of total products and a very small
yield of polycarboxylic acids. These differences are probably due to the smaller
ring structures and the larger concentration of phenolic substituents in the sub-
bituminous coal. Both coals gave a little acetic acid, evidently from methyl
substituents, considerable malonic acid, presumably from methylene bridges, and
considerable succinic acid. The latter acid could be due to either ethylene
bridges (diarylethane structures) or 9,lO-dihydrophenanthrene. It should be noted
that the compounds identified in the oxidation products of tetralin are essentially
absent from both coals indicating negligible participation of tetralin type struc-
tures. The ~ e s u l t s obtained by this technique are intriguing but additional model
compound work is needed to reach definitive conclusions.
2.1.3 Oxygen functionalities
Oxygen plays a pervasive role in almost all aspects of coal reactivity. Nitro-
gen, in the form of basic groups, is involved in the secondary intermolecular
forces in coal or coal derived liquids. It is also the source of much of the NO
x
generated in coal combustion. Sulfur is of concern primarily as a pollutant
precursor.
Oxygen is present as phenolic hydroxyl and carboxylic acid groups, aryl-aryl
or alkyl-aryl ether bridges, and ring oxygen, chiefly in furan-type structures.
Phenolic hydroxyls are evident in infrared spectra (refs. 15,16) and have been
7
TABLE 2.1
Products from the oxidation of pure compounds with aqueous H
2
0
2
+CF
3
COOH (refs. 12-14).
Yields of carboxylic acids (weight %)
Compound oxidized
Acetic Propionic Malonic Succinic
Other
to1uene 87
ethyl benzene 19
propyl benzene 5
2-methylnaphthalene 83
diphenylmethane
diphenylethane
tetralin
o
71
8
9
9 73
14
31
butyric: 73
cyclohexene-l,2-
dicarboxylic anhydride:
71
benzenedicarbo-
xyl ic acid: 70
benzenedicarboxylic
acid: 9, other
benzenemono-, di-,
tri- and tetracarbo-
xylic acids: 200
a
Illinois #6 l.4
b
2.4
b
7.4
b
same as above: 30
b
bituminous
Wyodak lo8
c
11 .4
c
4.9
c
same as above: 7
c
subbitumi nous
a, b, c relative yields
8
determined by a technique based on the formation of trimethylsilyl
ethers (ref. 17). They constitute one to two thirds of the oxygen in coal. The
determination of ether bridges which occur in smaller amounts has been indirect
and less reliable. Some investigators (e.g. ref. 18) have suggested that ether
bridges are the predominant links between the structural units of bituminous
coals. However, the bulk of the evidence is that aliphatic bridges are at least
as abundant. Carboxylic groups are found in considerable concentration in sub-
bituminous coals and lignites. Furan-type ring oxygen comprises a small fraction
of the total and is relatively unimportant from the standpoint of reactivity. A
quantitative determination of oxygen functional groups in several coals has been
made in connection to liquefaction studies (ref. 19).
The functional groups discussed above are organized in modular units (or simply
units) which are covalently linked to coal molecules. The coal molecule in Fig.
2.2 is shown to illustrate terminology rather than to suggest a representative
structure. vie do not wish to join in the debate surrounding "coal models". A
modular unit, e.g. unit A, consists of an nucleus, in this case naphtha-
lene, aliphatic chains, mostly methyl or ethyl, phenolic hydroxyl andcarboxylic
acid substitutents. Some units such as unit B contain a partially hydrogenated
nucleus, in this case the 9,lO-dihydrophenanthrene. The modular units are linked
by bridges, mostly methylene, ethylene and ether. The term bridge will denote a
link that can be thermally broken. The unit C, for example, contains a biphenyl
nucleus. The bond between the two benzene rings is not counted as a bridge be-
cause of its high dissociation energy.
Fig. 2.2. Illustration of modular units, nuclei, bridges and chains in a coal
molecule.
9
It has been indicated earlier that the most abundant nuclei in subbituminous
and bituminous coals consist of two or three condensed rings each, as illustrated
in Fig. 2.2. It may be further inquired what is the number of modular units or
the molecular weight of the coal molecule. A seasoned discussion of the evidence
concerning molecular weights was presented by Given (ref. 2). The number average
molecular weight of pyridine extracts of coal and of a Li/amine reduced coal (re-
ducing aromatic rings but probably not cleaving bonds) were in the range 1,000 to
4,000. The extracts of the reduced coal represented 40-80% of the original coal
by weight, hence the molecular weight of 1,000-4,000 can be considered fairly
representative. The remalnlng insoluble material may have higher molecular
weight or higher concentration of polar groups.
Assuming a molecular weight of the modular unit of about 200-250, the coal
molecule should contain 5 to 20 modular units. The arrangement of these units
is unlikely to be linear, because that would imply a number of bridges much lower
than inferred from devolatilization experiments. A two-dimensional structure
comprising sheets is certainly possible. The coal molecules are held together
by secondary (non-covalent) bonds like hydrogen bonds or charge transfer acid-
base complexes. Such secondary bonds are severed during extraction by pyridine
or other good solvents or at elevated temperatures.
2.2 STRUCTURAL AND FUNCTIONAL GROUP ANALYSIS
The analytical data that are commonly available for coal derived liquids -
extracts, pyrolyzates, and products of hydrogenation - are elemental analysis,
and lH nmr spectra and, less routinely, 13C spectra. These data can be trans-
lated to a form easy to visualize and correlate with reactivity properties by
certain procedures that have become known as structural analysis and functional
group analysis. The more common method of structural analysis seeks to determine
features such as the size of the aromatic nuclei, the degree of substitution and
the relative abundance of various types of aliphatic chains. If the molecular
weight is also specified, perhaps in some narrow range, representative or "aver-
age" molecular formulas can be constructed compatible with the data. The related
term functional group analysis indicates the determination of the concentration
of specific funtional groups without being concerned with representative molecular
formulas.
2.2.1 The Grown and Ladner method (refs. 20,21)
This simplest and best established of the structural analysis procedures util-
izes the atomic ratios C/H, O/H, and the hydroqen distribution H + frl, H /H,
~ ar 0 a
He/H, i.e. the fractions of aromatic + phenolic hydrogen, alpha hydrogen and beta
or further removed from the aromatic ring hydrogen. The following assumptions
or specifications are made:
10
(i) A fraction (usually 60%) of the oxygen is assumed to be phenolic
in accordance with wet chemical studies indicating a phenolic fraction of oxygen
in the range 0.4 - 0.6.
(ii) The remaining oxygen (other than phenolic) is assumed to be connected to
aromatic carbons as substituent (e.g. quinonic structures or methoxy groups),
and not as ring oxygen.
(iii) The ratios x = H /C , y = HS/C
S
are specified within reasonable limits.
. ex ex
The values x = y = 2 have been used in the paper of Brown and Ladner (ref. 21).
(iv) Biphenyl bonds are assumed absent.
(v) Sulfur and nitrogen groups are ignored.
Subtracting the phenolic hydrogen (assumption ii) from H + provides the aromatic
ar ('
hydrogen, Har/H. The aromaticity fC/" i.e. the of atomatic carbon,
then calculated with the aid of assumption (iii),
C-Cex-C
S
(C/H)-(Hex/XH)-(HS/YH)
f = C (C/H)
The next quantity of interest is the degree of substitution o,which is the ratio
of substituted aromatic carbons to the total number of peripheral aromatic car-
bons. This is calculated using assumptions (iv) and (v):
o =
C +0
ex
C +O+H
ex ar
= (H /xH)+(O/H)+(H /H)
ex ar
Finally, if H is the number of aromatic hydrogen and other aromatic substitu-
aru
ents and Car the number of aromatic carbons, the quantity Haru/Car which repre-
sents the ratio of Hand C in the hypothetical unsubstituted hydrocarbon, is
given by:
H
aru
_
H + Cex + 0
ar
-C- -
C
ar ar
(Ha/H) + (Hc/xH) + (O/H)
(f C/H)
ex
In addition to its indirect determination given above, the aromaticity can be
obtained directly from 13C magnetic resonance spectra. The two values were found
in very good agreement in a study of various coal derived liquids (ref. 22).
The ratio H /C is related to the size of the aromatic nucleus. For example
aru ar
naphthalene, anthracene and phenanthrene have Haru/Car values of 0.8, 0.714 re-
spectively. A computed value of 0.75 might thus indicate a mixture largely con-
sisting of naphthalene and phenanthrene or anthracene. This interpretation of
H /C is obscured, however, by the presence of heteroatoms, biphenyl bonds
aru ar
and hydroaromatic nuclei such as 9,10-dihydrophenanthrene.
2.2.2 Detailed structural analysis (refs. 23,24)
Bartle and Smith (ref. 23) developed a detailed characterization of the aliph-
atic structures based on the ability to break down the IH nmr spectrum into the
following types of hydrogen: H
ar
+
o
' hydrogen in methylene connecting two rings
11
(e.g. a methylene bridge); hydrogen of methylene in acenaphthene or indene as
well as methylene connected to rings in highly sterically hindered positions;
H
a
not belonging to the two previous categories; hydrogen in -CH
2
-, -CH- groups
S or further to a ring + CH
3
S to a ring; hydrogen of CH
3
r or further to a ring.
These data are supplemented by the following assumptions:
(i) A certain fraction of oxygen is attributed to phenolic groups; the remain-
ing is assumed to be in dibenzofurans.
(ii) Nitrogen is assumed to be in rings of the type of pyridine and carbazole,
while sulfur is assumed to be in dibenzothiophene rings.
(iii) The aliphatic chains considered include methyl, ethyl, propyl, isopropyl,
n-butyl, isobutyl and sec-but.yl,. The rati 0 of n-butyl to n-propyl is gi ven some
convenient value. The n u m ~ 2 ~ s of n-butyl, i-butyl and sec-butyl are taken to be
eqlJal.
Using assumptions (i) and (ii)the heteroatoms are replaced with C and H to
produce an equivalent structure containing C and Honly. Balances are then writ-
ten for H
a
, H
S
and H which in conjunction with assumption (iii) yield the con-
centrations of the various aliphatic chains.
The above procedure has been used to determine representative structures for
narrow molecular weight fractions of a supercritical coal extract (ref. 8).
For example, the structure below was proposed for the methanol eluate of the ex-
tract:
The method was recently modified and further applied to characterize coal extracts
(ref. 24). By combining the information from IH and 13C nmr spectra, it
was possible to largely dispense with the assumptions (i)-(iii) above in an
application to petroleum fractions (ref. 25). The fundamentals of IH and 13C nmr
spectroscopy as it pertains to complex mixtures, coal liquids in particular, has
been recently reviewed by Bartle and Jones (ref. 26).
2.2.3 Computer-assisted molecular structure construction (ref. 27)
The structural parameters and the concentrations of functional groups derived
by the methods of refs. 25-26 can be assembled into average or representative
molecular formulas, assuming suitab1e molecular weight information. In the afore-
mentioned references the construction of molecular formulas was carried out by
trial and error. Since this procedure becomes impractical with increasing molec-
ular weight, Oka et al.(ref. 27) developed a computer-based technique for generat-
12
ing molecular formulas from data on lH nmr, elemental analysis and molecular
wei ght.
the linear programming formulation it is required to maximize a
(2.1)
i = l, ... ,m
2.2.4 Functional group analysis by linear programming
As already discussed, structural analysis can be carried out at two levels.
At the first level, it is sought to derive various features such as fa or H
aru
/
C representing averages for the mixture examined. The concentrations of vari-
ar .
ous aliphatic groups can also be estimated if the lH nmr data are sufficiently
detailed. At the second level, a representative or average moelecular formula
is sought, compatible with the available data. Such a formula is illustrative
of the type of possible structures but cannot convey the variety of different
molecules that may be present even in a narrow fraction of the sample.
The thermal reactions of pyrolysis, hydropyrolysis and liquefaction can be
largely understood in terms of functional groups rather than molecules, therefore
the first level of structural analysis is usually sufficient. The problem can
then be stated as follows: Given elemental analysis, nmr and possibly other data
calculate the concentrations of a postulated set of functional groups. Invariably,
the number of unknowns exceeds the number of balance equations incorporating the
data, therefore the problem accepts an infinite number of solutions. To obtain a
unique solution, additional assumptions in the form of values for the ratios of
some of the unknowns may be introduced as in refs. 23-25. This is a convenient
device but in some cases the specified values may lead to improbable, even nega-
tive values for some concentrations. Thus, some trial and error is normally re-
quired in the choice of the assumed values.
A more systematic procedure for generating meaningful solutions of the undeter-
mined problem of functional group analysis is afforded by linear programming.
Suppose that Y1 ""'Yn are the unknown concentrations of the selected functional
groups and let the balance equations have the form
n
2: A.. y. = b.
j=llJ J 1
where m<n. In
(2.2)
+ qnYn
Eq. (1.1) and the obvious constraint y.>O. In this notation A.. are
1- 1J
stoichiometric coefficients, b
i
are conserved quantities, directly derived from
the data, and qi are positive numbers chosen on the basis of experience and intui-
tion. Suppose, for example, that Y1'Y2 represent the concentrations of methylene
and ethylene bridges. A complete lack of preference as to which of the two is
more abundant can be expressed by setting q2=2ql' the factor 2 deriving from the
two-to-one carbon between the two bridges. To express an intuitive preference
quantity
J = qlY 1 +
subject to
13
for methylene bridges, on the other hand, one might set q2 = 0.2q 1
For purposes of illustration this technique was applied to a product of lique-
faction of a bituminous coal with the analytical data given in the following
tab1e.
TABLE 2.2
Data for a donor solvent extract of a bituminous coal (ref. 28).
Conserved quantity
Concentration
_ 2
10 g-mols/g extract
Conserved quantity
Concentration
L
10 g-mols/g extract
C b
l
5.31
H
S
+
b
6
0.52
ar
Cal
b
2
1.86 0 b
7
0.28
H
ar
+
o
b
3
2.14 S b
8
0.16
H b
4
2.21 N b
g
0.14
a
H
S
b
5
1. 62
The functional groups used in this illustration, twenty-one in number, are
listed in Fig. 2.3. The coefficients A
ij
are obtained by inspection of Table 2.2
and Fig. 2.3. A
ij
is the contribution of functional group j (Fig. 2.3) to the
conserved quantity i (Table 2.2). For example, A
2
,J3 = 2, A
3
,13 = -1 because a
chain -CH
2
CH
3
replaces a group H
ar
+
o
which would otherwise be attached to the
aromatic carbon. Similarly, A
3
,17 = -2, A
S
,15 = -1 because the group o-CH
3
re-
places a H
S
' Table 2.4 later gives the complete matrix of coefficients A
ij
.
As mentioned below, the choice of the coefficients q. is a statement of experi-
1
ence or intuition concerning the relative abundance of various functional groups.
For example ql = 1, q2 = 5, q3 = 5, q4 = 1, q5 = 1 states that it is more prob-
able that the mixture contains groups 2 and 3 than groups 1, 4 and 5. The re-
sulting solution to the problem may still contain groups 1, 4 and 5, however.
Even when qj = 0, the solution may still contain yj>O. Each q-set provides a
solution satisfying equations (2.1). Different q-sets do not necessarily pro-
duce different solutions. If Yl ""'Yn and Y1 , ... y ~ are the solutions corres-
ponding to ql , ... ,qn and q;, .. q ~ then the linear combination yj = AYj+(l-A)yj
( O ~ A ~ l ) is also a solution of equation (2.1) although it does not maximize any
particular J. However, it is perfectly acceptable from our standpoint. The util-
ization of the maximization problem is mainly a device to obtain feasible solu-
tions.
14
COO
cOO?
f
OO?
00
2 3 4 5
C
~ o - -OH-
6 7 8
D o

5 5
9 10 II
-CH
3
-CH
2
CH, ~ C H 2 C H 2 C H , e-- CH, o-CH,
12 13 14 15 16
-CH
2
- ~ C H 2 C H 2 -
0 CO
17 18 19 20 21
Fig. 2.3. Functional groups used in the analysis of the data of Table 2.2.
The symbols ., @, 0 denote attachment to aromatic, a and S carbons, respectively.
Table 2.4 lists eleven distinct q-sets chosen to emphasize different function-
al groups. For example, sets 3 and 4 differ only in the coefficients q2' with
set 3 emphasizing the phenathrene ring while set 2 placing equal emphasis on
naphthalene and phenanthrene. Table 2.5 gives the results of calculations with
each of the q-sets of Table 2.4. Several features of these results should be
noted:
(i) For any given set of qj' the number of non-zero Yj is equal to the number
of data points, m. The remaining Yj are zero. By combining such solutions it
is possible to generate feasible solutions with a larger number of positive com-
ponents. All the q-sets of Table 2.4 produce zero values for the components Y4'
Y5' Y
6
, Y1
0
' Y14' Y20' Y2l without precluding, however, that some other q-set may
produce non-zero values for some of these components.
(ii) A comparison of the solutions corresponding to the q-sets 5 and 6 differing
only in ql' q2' and q3 shows set 5 producing only benzene and set 6 producing
only naphthalene. The transition from benzene to naphthalene is accompanied by
a reduction in the number of biphenyl bridges.
16
(iii) The concentrations of the hydroaromatic groups 20,21 are zero for all q-sets.
It should be kept in mind, however, that in the functional group analysis a struc-
ture 1ike
is regarded as containing two methylene bridges rather than a hydroaromatic group.
(iv) The q-sets with q7 = qs yield concentrations Y7 = 0, yS O. After suffic-
iently increasing q7 the concentrations become Y7 > yS = 0 with a simultaneous
change in the functional groups 17,lS and 19 representing bridges. A solution with
Y7 > 0, ys > 0 can be obtained by combining the solutions of q-sets 6 and 7 using
equal weights. This solution has 11 positive components and is listed under col-
umn C of Table 2.5.
17
TABLE 2.4
Sets of coefficients
qi
Set. No.
2 3 4 5 6 7 8 9 10 11
qj
2
2
q2 2 5 2 5 5
q3 2 5 5 2 5 5 5 20 20 20
q4
qs
q"
2 2 2 2 2 2 2
.)
')
2 L
q7 4 4 4 4 4 S 5 4 4 4
qg
4 4 4 4 4 2 2 4 4 4
qa 5 5 5 5 5 5 5 5 5 5
ql0 5 5 5 5 5 5 5 5 5 5
q11 10 10 10 10 10 10 10 10 10 10
q12 2 2 2 5 5 5 5 2 2 2
q13 2 2 2 5 5 5 5 2 2 5
q14 2 2 2 5 5 5 5 2 2 2
q15 2 2 2 5 5 5 5 2 2 2
q]6 2 2 2 5 5 5 5 2 2 2
q17 2 2 2 2 2 2 2 2
q18 4 4 4 2 2 2 2 4
q19 2 2 2 2 2 2 2 2 4 4
q20 5 5 5 5 5 5 5 5 5 5
q21 5 5 5 5 5 5 5 5 5 5
18
TABLE 2.5
2
Concentrations of functional groups (la-g-mols!gr extract) computed by
linear programming for various q sets.
q-set
Group
No. 2 3 4 5 6 7 8 9 C
6.59 6.59 a 6.59 6.59 a a 6.59 a a
2 a a J.95 a a 3.95 3.95 a a 3.95
3 0 0 a a a a a a 2.82 a
4 a a a a a a a a a a
5 a a a a a a a a a a
6 a a a a a a a a a a
7 a a a a a a 2.82 2.82 a 1.41
8 2.82 2.82 2.82 2.82 2.82 2.82 a a 2.82 1.41
9 a.16 a.16 a.16 a.16 a.16 a.16 a.16 a.16 a.16 a.16
la a a a a a a a a a a
11 1. 36 1. 36 1.36 1. 36 1.36 1. 36 1.36 1.36 1. 36 1. 36
12 6.21 6.21 6.21 6.21 6.21 6.21 6.21 6.21 6.21 6.21
13 a a a a a a a a a a
14 a a a a a a a a a a
15 5.99 5.99 5.99 5.99 5.99 5.99 5.99 5.99 5.99 5.99
16 1. 73 1. 73 1.73 1.73 1. 73 1.73 1.73 1.73 1. 73 1. 73
17 6.48 8.84 2.53 6.48 6.48 2.53 a 3.66 a.83 1.26
18 4.71 a 4.71 4.71 4.71 4.71 4.13 4.71 4.71 4.42
19 a 2.36 a a a a a.29 a a a.15
2a a a a a a a a a a a
21 a a a a a a a a a a
C: average of results of q-sets 6 and 7.
19
Chapter 3
THERMAL REACTIONS OF COAL
This chapter deals with the rate parameters of the principal elementary reactions
of pyrolysis, hydropyrolysis and liquid-phase hydrogenation. While considerable
data is available in the literature about rate parameters ,in most cases recourse
must be made to estimation techniques such as group additivity and transition state
theory. These two techniques, group additivity in particular, have been largely
developed by Benson and coworkers (e.g. refs. 29,30) and are popularly known as
thermochemical kinetics. Much of this chapter is taken up by the application of
thermochemical kinetics to various elementary reactions of interest. The estimated
parameters along with those taken from the literature are compiled in a table at
the end of the chapter.
One of the uncertainties associated with estimating rate parameters for coal
reactions is due to the condensed nature of the coal phase which is the reaction
medium. Almost all reported rate values and group contribution information refer
to gas phase kinetics. Another uncertainty arises from the fact that coal molecules
contain, primarily, condensed ring aromatics while much of the reported experi-
mental information refers to single ring compounds. When these uncertainties are
added to the significant uncertainty associated with the structure of the transi-
tion state and the frequencies of its vibrational modes it will be realized that
estimated A-factors could be easily in error by a factor of ten. Nevertheless,
the estimates discussed here are very useful in the absence of other information
and can obviously be improved by comparison with data from model compound studies.
With these difficulties in mind we now proceed to examine several classes of
reactions: bond dissociation, hydrogen abstraction and hydrogen addition. These
are all free radical reactions. Reactions other than free radical, e.g. concerted
reactions, have only recently been examined relative to coal pyrolysis and will
be briefly surveyed in the last section.
3. I BOND DISSOCIATION WITH PRODUCTION OF TWO RADICALS
The following reactions will be considered with Ph, Ph' representing aromatic
nuclei such as naphthalene, phenanthrene, etc.
Ph-CH
3
~ Ph-CH
Z
+ H'
~ h - C H 2 C H 3 ~ P h - C H ~ + C H ~
Ph-CH
2
CH
2
CH
3
~ PhCH
Z
+ CH
3
CH
Z
Ph-CH
2
-Ph' ~ Ph-CH
Z
+ Ph'
Ph-CH
2
CH
2
-Ph' ~ Ph-CH
Z
+ Ph'-CH
2

Ph-0-CH
3
~ Ph-D' + CH
3
'
(D1)
(D2)
(D3)
(D4)
(D5)
(D6)
20
Ph-a-Ph' ~ Ph-a + Ph'
Ph-0-CH
2
-Ph' ~ Ph-a + Ph'-CH
2
'
( 07)
(08)
The rate of each of these elementary reactions can be expressed in the Arrhenius
form
k = Aexp( -E/RT)
where A is the "A-factor" and E the experimental activation energy, both being
functions of temperature. The activation energy E can be estimated using the
fact that the reverse reaction, free radical recombination, has very small acti-
vation energy which is conventionally taken as zero. Thus
E = LlH (3.1 )
where LlH is the standard heat (or enthalpy) of reaction, at the temperature of
interest and pressure of one atmosphere. LlH is also somewhat less accurately
known as the bond dissociation "energy". The heat of reaction LlH can be estimated
by the group additivity method as shown in the examples below.
The A-factor is given by transition state theory as
k T
A = e - - - ~ - - - exp(LlSot/R) (3.2)
where Lls
ot
is the standard (atmospheric pressure) activation entropy, i.e. the
difference in entropy between the transition state (or activated complex) and
the reactants. The activation entropy can be estimated by the methods of statis-
tical mechanics if the structure of the activated complex is known. This estima-
tion involves a great deal of uncertainty because of lack of reliable information
about several of the vibrational frequencies of the activated complex.
Benson and O'Neal (ref. 29) have evaluated experimental data for several of the
reactions (01)-(08) and have recommended preferred parameter values in each case.
In the following subsections we shall illustrate the theoretical estimation tech-
niques on reactions which are not included in the compedium of Benson and O'Neal.
The estimated parameters along with values from the literature are listed in the
table at the end of the chapter.
Examples 1 to 5 treat the estimation of activation energies while examples 6 to
8 the estimation of A-factors.
3.1.1 Activation energies
Example 1.
In this example we examine reaction 01 with Ph an unsubstituted phenyl. The
calculations will be initially made for 300
0
Ksince most of the available data
refer to that temperature. Subsequently, the estimate will be revised to apply
to 800
o
K. The heat of reaction can be expressed in terms of heats of formation as
LlH = LlfH(H) + LlfH(Ph-CH
Z
) - Ll
f
H(Ph-CH
3
)
The last two terms can be estimated by group additivity as follows:
21
6
f
H(Ph-CH
Z
) = 5[C
B
-(H)] + [CB-(C)] + [Co-(C
B
)(H)2]
6
f
H(Ph-CH
3
) = 5[C
B
-(H)] + [CB-(C)] + [C-(C
B
)(H)3]
where the brackets denote heats of formation and the groups within the brackets
are written with the notation of ref. 30. The tables in ref. 30 give [CB-(C')]
[CB-(C)], hence
6H = [Ho] + [C'-(C
B
)(H)2] - [C-(C
B
)(H)3] = 52.1 + 23.0 - (-10.2) 85.3 kcal/g-mol.
For comparison purposes we examine the aliphatic analog,
CH
3
CH
3
7 CH
3
CH
z
+ H'
for which
6H = [H] + [C-(C)(H)2] - [C-(C)(H)3] = 98.1
The difference between the two energies
[C-(C)(H)2] - [C-(C
B
)(H)2] = 12.8 kcal/g-mol
is due to the interaction of the free electron with the n-bonding orbital of the
benzene ring and is called the resonance stabilization energy (RSE). The RSE is
responsible for the relatively low value of several bond dissociation energies
in coal. Table 3.1 lists the RSE of a-radicals deriving from larger rings as
estimated by Stein et al. (ref. 31).
TABLE 3.1 Resonance stabilization energies (kcal/g-mol) for radical with CH
2

at position A,B,C,D,E (ref. 31).

Compound

C A
00J8
B
A
13.0
16.4
18.5
B
14.7
15.3
C
22.2
o E
C
o
15.0 15.0
14.0 16.0 16.0
E
(05 ')
(05a)
Table 3.1
22
Using the RS values of Table 3.1 we can estimate the heat of various dissocia-
tion reactions as shown in the following examples. For convenience we shall use
the notation Ph
l
= phenyl, Ph
2
= naphthyl, Ph
3
= phenanthryl. Furthermore, P h ~
will denote 2-naphthyl, Phl will denote l-phenanthryl, etc.
Example 3.2
We consider the reaction
2 Z
Ph
2
-CH
2
CH
3
+ PhZ-CH
Z
' + CH
3
(OZ')
and its aliphatic analog
CH
3
CH
Z
CH
3
+ CH
3
CH
Z
' + CH
3
' (OZa)
The heat of (OZa) is well established (ref. 30) as 85, hence using the RSE value
of 14.9, we estimate for reaction (OZ'), 6H = 85-14.9 = 70.1.
Example 3.3
To estimate the heat of reaction
Z Z
PhZ-CHZ-PH
l
+ PhZ-CH
Z
' + Ph
l
(04')
we start from the simpler
Phl-CHZ-Ph
l
+ Phl-CH
Z
' + Ph
l
' (04")
for which 6H can be computed by group additivity using the values of ref. 30:
6H(04")= [Phl-CH
Z
'] + [Ph
l
] - 10[CB-(C
B
)] - 2[C
B
-(C)] - [C-(CB)z] = 45 + 78.5
- lOx 3.3 - Z x 5.51 - (-4.86) = 84.3.
The difference between the heats of reactions (04') and ( O ~ ' ) is equal to the
difference of the corresponding RSE which is 3.6, therefore 6H(04') = 84.3 - 3.6
= 80.7.
Because of this large value of 6H, the rate of direct dissociation of methylene
bridges is negligible relative to other pyrolysis reactions. However, two indirect
dissociation mechanisms are much more energetically favorable. The first proceeds
by the addition of hydrogen atoms and other small radicals to the aromatic ring.
The second is operative in the presence of phenolic hydroxyl groups in the ortho
and para position. These two mechanisms will be discussed in following subsections.
Example 3.4
Here we consider reaction (05) and its aliphatic analog,
1 Z 1 Z
Ph
3
-CH
Z
CH
Z
-Ph
Z
+ Ph
3
-CH
Z
' + Phz-CH
Z
'
CH
3
CH
Z
CH
Z
CH
3
+ CH
3
CH
Z
' + CH
3
CH
Z
'
The heat of (05a) is 8Z, so that using the appropriate RSE values from
we obtain 6H(05') = 8Z - 15.6 - 16.6 = 48.9.
Example 3.5
We next estimate the heats of reactions (06)-(08) which involve the dissociation
of carbon-oxygen bonds. Because the heats of formation of the phenoxy radicals
(Ph-O') are not available we will assume that their resonance stabilization energies
are the same as those of the corresponding benzyl radicals,
[O'-(C)] - [O'-(C
B
)] = [C'-(C)(H)Z] - [C'-(CB)(H)Z] = lZ.8
23
The value of [O_(C)] can be calculated from the heat of formation of CH
3
CH
2
0',
6
f
H[CH
3
CH
2
0] = [O_(C)] + [C-(C)(H)2(0.)] + [C-(C)(H)3]
In this equation 6
f
H[CH
3
CH
2
0'], [C-(C)(H)3] are listed in reference 30, while
[C-(C)(H)2(0')] may be approximated by [C-(C)(H)2(O)]. The result is [O'-(C)]
= 14.3 so that [O'-(C
B
)] = 14.3-12.8 = 1.5. Using this value in group additivity
calculations we find the heats of reactions (06),(07),(08) to be 68.3, 85.5, 55.6
which vary by no more than +3 kcal from the heats of reactions (02),(03), and
(04), respectively.
Since the aromatic nuclei in coal are substituted to a considerable degree we
must consider the effect of substituents on the heats of the bond dissociation
reactions. Barton and Stein (ref. 32) conducted some model compound studies show-
ing that the heat of reaction (02) is reduced by about 1 to 2 kcal/g-mol by an
ortho-methyl substituent. Meta and para substituents were found to have little
or no effect. Earlier studies summarized in reference 32 had shown that meta sub-
stitution by a methyl group has no effect, para has a sliqht effect and ortho has
a more significant effect, up to 3 kcal/g-mol. An investigation of the pyrolysis
of cresols (ref. 33) showed that o-cresol reacts faster than m-cresol by a
factor of about three at 806K. In both cases the rate determining step was the
dissociation of a benzylic hydrogen. Assuming equal A-factors, the activation
energy in the ortho-compound must be lower than that of the meta-compound by about
2 kcal/g-mol.
All the above calculations refer to a temperature of 300
o
K. To calculate the
heats of reaction at higher temperatures, ego 800
o
K, requires the pertinent heat
capacities. These can be computed by group additivity using the data of refer-
ence 30. Using the symbol [ ]c to denote the heat capacity of a illus-
trate the calculation for the reaction of example 3.2 above.
6C
p
(3000K) = [CH
3
']c + [C.-(C
B
)(H)2]c - [C-(C)(H)3]c - [C-(C
B
)(C)(H)2]c =
3.26 cal/g-mol oK.
The corresponding values at 400,500,600 and 800
0
1( are 2.07,1.32,0.62,0.03. Using
these values we obtain
6H(8000K) - 6H(300oK) (T)dt 0>0.58 kcal/g-mol
300 P
A similar calculation for reaction gives 6H(8000K) - 6H(300oK) 0.08. These
differences turn out to be quite small.
3.1.2 A-factors
The A-factor is given in terms of the entropy of activation by Eq.(3.2). In
gas phase kinetics 6so
t
can be broken down to contributions from various degrees
of freedom which can be estimated by the techniques of ref. 30 as shown in several
examples below. However, this procedure cannot be rigorously applied to coal pyrol-
ysis because the reaction medium is an amorphous solid or a viscous liquid, there-
fore the partition function cannot be factored out as in the ideal gas case. For
24
certain reactions that can take place in solution as well as in the gas phase it
has been found experimentally that as long as the solvent does not play any chem-
ical role, the A-factor in solution was comparable to the A-factor in the gas
phase (ref. 34). In a recent study (ref. 35) the rate constants of reactions
(05) and (08) were estimated to be 0.42 and 1.14 in the liquid and gas phases re-
spectively. However, the experimental data base for these estimates was limited.
Moreover, it is not known whether the rough equivalence of rate constants also
aPplies to a condensed phase which is an amorphous solid or a very viscous liquid.
In the absence of better information it can still be assumed that A-factors com-
puted for gas phase kinetics are roughly applicable to the coal phase at least
when the reactions are not diffusion limited.
In carrying out the detailed calculation of the entropy of activation it must
be kept in mind that the reacting functional group is actually attached to a rather
bulky molecule. For example, in reaction (OZ) Ph is an aromatic nucleus such as
naphthalene bridged to other similar units. A more descriptive notation for this
reaction then would be
U1-Ph-CH
Z
CH
3
-r CM-Ph-CH
Z
" + CH
3

where Ph is connected by one or more bridges to other sections of the coal molecule
denoted here as CM. The reactant coal molecule may have molecular weight from one
to five thousand or more in the case of pyrolysis but less in the case of hydropy-
rolysis and liquefaction. With such large molecular weights the differences be-
tween the reactant molecule and the transition state
01-Ph-CH
2
'"' CH
3
are small as far as rotational partition functions are concerned. The external
symmetry is likewise the same, one, in both reactant and transition state. As a
result of the large size and complex nature of the molecule, the main contributions
to the entropy of activation come from vibrations, mainly bending modes, and in-
ternal rotations.
A persistent difficulty in estimating A-factors for condensed phase reactions
is the possibility of diffusional limitations. This difficulty is especially pro-
nounced in coal pyrolysis because of the high viscosity of the condensed phase.
Consider for example reaction (05) written in the more detailed form
k] k;
Ph-CH
Z
' -:.. Ph-CH
Z
+
k] k
2
where the bar indicates a couple of radicals in close proximity. Because of their
bulky size and the high viscosity of the coal phase the two radicals may recombine
before diffusing sufficiently apart from each other. This "cage" effect can be
analyzed by applying the steady state approximation to the intermediate configur-
ation to obtain for the net dissociation rate the expression
k)k
2
k;k
2
r = -- -
k;+k; k;+k;
25
under diffusion limited conditions k{ k
2
and the above expression simplifies to
r = - kzlPh-CH
Z
-CH
Z
]
where K
1
= k1/k{. The diffusion parameters k
1
, k
2
have been calculated theoretic-
ally for the case of polymerization reactions (refs. 36-38), In the case of py-
rolysis such theoretical calculations are not feasible. We will return to this
subject in the chapter dealing with kinetic modeling.
We shall now present examples of estimation of A-factors for several of the
d1ssociation reactions (01)-(08). It should be emphasized once more that the
assignment of frequencies to internal rotation and bending modes is subject to a
great deal of uncertainty. This is the main reason for the fact that theoretically
calculated A-factors can be easily in error by a factor of ten and in some cases
more. If on the other hand it is possible to exploit analogies with experimentally
known A-factors much more reliable estimates Cctn be cbtained. The examples below
illustrate the adjustments required in exploiting analogies. The various
frequencies are assigned using the data of ref. 30.
Example 3.6
In this example we will estimate the A-factor of reaction (OZ). The starting
point is the reaction with Ph an unattached phenyl, PhI' for which the experimental
value logA = 15.3 at 1,000oK is given in ref. Z9. The estimation of the entropy
of activation is based on the change between reactant and transition state,
Ph
l
-CH
Z
CH
3
-;. Ph
l
-CH
Z
.... G1
3
(dZ)
where the dots indicate the partially broken bond. The contribution of the vari-
ous degrees of freedom are as follows:
(i) Translation, spin and symmetry make no contribution as being the same in
reactant and transition state.
(ii) Two principal moments of inertia are increased by a factor of 1.8 with a
Rn 1.8 = 1.Z rotational contribution.
(iii) A C-C stretch (1,000 cm-
I
) becomes the reaction coordinate contributing -1.4.
(iv) The internal rotation about the Phl-C bond (barrier changes from Z to 15
kcal) contributes -1.8; the internal rotation about the CH
Z
-CH
3
bond (barrier
changes from 3 to a kcal) contributes O.Z.
(v) There are four bending modes about the CH
Z
-CH
3
bond. The frequencies of
these modes in the reactant can be estimated by analogy to the aliphatic dnalog
CH
3
-CH
3
for which ref. 30 gives 1,000 cm-
I
Since the reduced mass of Ph-CH
Z
CH
3
is about two times larger, the associated frequencies are about 700 cm-
I
The
corresponding frequencies in the transition state can be used as adjustable param-
eters to bring the estimated total 6so
t
in accord with the experimental value. The
required frequencies turn out to be ZlO cm-
I
(d
5
)
= 3.0 at 10000K. An analysis of
for the contribution of the bending
(d4)
-1 .
as in the previous example 170 cm ln
state making a contribution of 4 x 1.5
26
We now consider the case when Ph is part of a more bulky coal molecule with
molecular weight about 500,
CM-Ph-CH
Z
CH
3
+ CM-Ph-CH
2
CH
3

The only difference with (dZ) is that the contribution of rotation is now negligible.
The bending frequencies remain the same because the reduced mass changes by the
same factor in the reactant and the transition state. At 10000K, 6so
t
= 5.7 - 1.2
= 4.5 and log A = 15.0. Repeating the calculations at 800
0
Kwe obtain 6so
t
= 6.4,
log A = 15. O.
Example 3.7
To estimate the A-factor for reaction (05) we start with the base case
Ph1-CHZCHZ-Ph
1
+ Ph
1
-CH
2
CHZ'Ph
1
t
for which ref. Z9 gives log = 14.4 or 6SO
6s
ot
can provide as h2fore a crude estimate
modes:
(i) Translation, spin and symmetry do not change.
(ii) Rotation: two moments increase by a factor of two contributing RnZ = 1.4.
(iii) A C-C stretch becomes the reaction coordinate contributing - 1.4 units.
(iv) Two internal rotations about the Ph1-C bonds (barrier changes from Z to 15
kcal) contribute 2 x (-1.8) = -3.6 units; an internal rotation about the CH
Z
-CH
2
bond becomes a free rotation resulting in an entropy change of 0.2 units.
(v) To match the experimental 6so
t
= 3 an additional 6.4 units are needed, which
can be assigned to four bending modes. If the frequency of each of these bonds in
the reactant is 400 cm-
1
, the frequency in the transition state must be about 180 cm-
1
We now consider the reaction
(d)
differing from d
5
mainly in the molecular weight of the group CM-Ph-CH
Z
' If this
weight is taken as 500 the bending frequencies are 400 = 170 cm-
1
180 = 77 cm-
1
. Using these frequencies and recognizing as before that
rotation makes a negligible contribution we find that for 800
0
K, 6so
t
= 1.1 and
log A = 13.9.
Example 3.8
We now go back to reaction (04),
CM-Ph-CHZ-Ph-CM +
where we distinguish the following contributions at 800
0
K,
(i) Translation, spin and symmetry do not change.
(ii) The rotational change is negligible.
(iii) The internal rotations about the Ph-CH
2
bond and the bond contribute
-Z.l and 0.2 units, respectively.
(iv) The four bending modes will be taken
the reactant and 77 cm-
1
in the transition
27
6 units. The sum of these components is Lso
t
3.0 and log A = 14.3.
We have analyzed, rather crudely, the reactions (02), (04) and (05). The A-
factor for (01) is estimated by adjusting the experimental value of log A = 15.5
at 1000
0
K reported for the dissociation of toluene. The A-factors of (06)-(08)
can be assumed to be approximately eoual to those of (02)-(05) respectively. The
results of these estimates are all listed in the last table of the chapter.
3.1.3 The effect of phenolic hydroxyl groups
It was mentioned earlier that pehnolic hydroxyl groups have a profound effect
on the rates of dissociation reactions. The reactions under consideration are
HO-Ph-CH
2
-X + HO-Ph-CH
2
+ X
where the -OH is ortho or para to the benzylic carbon and X is one of the groups
H, CH
3
, Ph, CH
2
-Ph. In an early study (ref. 33) of cresol pyrolysis, o-cresol
and p-cresol were found to decompose about four times faster than m-cresol at
816K. The rate determining step in each case was the dissociation of a benzylic
Hatom. Assuming equal A-factors,the activation energy of the first two reactions
must have been about 3 kcal lower than that of the last reaction. Although not
examined in this study, the dissociation of the benzylic hydrogen in toluene and
m-cresol must have very similar activation energies.
The activating mechanism of ortho or para situated hydroxyl was recently identi-
fied as due to a keto-enol tautomerism (ref. 39),
..
Assuming the second step to be rate determining, the effective reaction rate con-
stant is k
2
k
1
/k
1
. The activation energy of step 2 was estimated as about only
45 kcal/g-mol while the equilibrium constant k
1
/k
1
was estimated as 10-
6
at 400C.
Assuming an A-factor equal to that of bibenzyl dissociation, the effective rate
constant k
2
k
1
/k
1
turns out to be several orders of magnitude larger than that of
direct dissociation.
(dbZ)
Z8
3.Z DISSOCIATION OF FREE RADICALS
The following are representative of this class of reactions:
Ph-CHCH
3
->- Ph-CH = CH
Z
+ H. (OB1)
Ph-tHCH
Z
CH
3
->- Ph-CH = CH
Z
+ CH
3
. (OBZ)
Ph-CH
Z
CH
Z
(HZ ->- Ph-CH
Z
+ CH
Z
= CH
Z
(OB3)
->- Ph-CH = + H (OB4)
Among the above reactions (OBI), (OBZ) and (OB4) involve the conversion of an
alpha radical to a higher energy radical. The attendant loss of resonance stabil-
ization energy results in a higher activation energy compared to the corresponding
aliphatic analog. For example reaction (OBZ) has an activation energy of 45 com-
pared to 34 for the aliphatic analog
CHpl
Z
tHCH
3
->- CH
3
+ CH
Z
= CHCH
3
The exception is reaction (OB3) where resonance stabilization energy is gained
and the activation energy is very low (9.6). The examples below focus on the
estimation of A-factors, using the same techniques as in section 3.1.Z. The estima-
tion of activation energies can be carried out in a straightforward way by group
additivity and is illustrated only in example 3.9.
Example 3.9
We start with reaction (OBI) and compare
CM-Ph-CHCH
3
->- (db1)
with the aliphatic analog
CH
3
-CHCH
3
->-
for which the experimental value log A = 14.3 at 500
0
K is reported (ref. Z9). The
two reactions differ mainly in the internal rotation, Ph-CH bond vs. CI1
3
-CH bond.
Taking this difference into account and adjusting for the temperature yields log A
15.1 at800oK.
Example 3.10
We next consider reaction (OBZ) . The heat of reaction can be immediately cal-
culated by group additivity using the group values of ref. 30 as I'IH=37.S. The
activation energy is given by E = 1'111 + vlhere is that of the reverse reaction.
The latter is about 7.Z, section 3.5, therefore E 45.
To estimate the A-factor we consider the entropy changes associated with
CM-Ph-CHCH CH ->- ... CII
Z 3 Z 3
and the aliphatic analog
CHiHCHZCH3 ->- CH
3
CW-'--'-'-CH
Z
CH
3

for which ref. Z9 gives the experimental value log A = 14.Z at 570
o
K. The differ-
ence in the entropies of (dbZ) and is mainly due to (i) rotation which
contributes to but not (dbZ) and (ii) internal rotation in (dbZ) is about
the Ph-C bond while in is about the CH
3
-C bond. The differences due to
(i) and (ii) nearly cancel each other so that for (dbZ), log A=14.Z at 570
0
Kand
log A=14.4 at SOooK.
(db3)
(db3')
29
Example 3.11
To estimate the A-factor of reaction (DG3) a comparison is made between
-+ Ct1-Ph-CH
2

CH CH CH tH CH CH .... CH -'---'--'--'-CH
322 Z-+ 3 Z Z Z
The predominant difference between the two activation entropies are due to (i)
rotation - which is negligible in (db3) - and (ii) bending modes about the weakened
CHZ-CH
Z
bond. In the absence of any information it is assumed that the contribution
of the bending modes to 6so
t
is the same in (db3) and (db3'). The difference due
to rotation is about 1.6 units, hence we set log A = 14.0 at 650
0
Kor 14.Z at 800
o
K.
Example 3.1Z
The last reaction (DB4) involves the entropy of
Ci.1-Ph-tHCH
Z
-Ph' -Ct1 -+
H"
consisting of the following parts:
(i) One C-H stretch becomes the reaction coordinate with a -0.1 units change.
(ii) One C-C stretch becomes a C-'---'--'--'-C stretch with a -0.5 change.
(iii) Two C-C-H bends become C-C ..H bends with Z x 0.4 = O.S change.
(iv) The barrier to internal rotation about the Ph-C bond changes from 15 to Z
kcal contributing Z.l units. The internal rotation about the t-c bond becomes a
torsion about the bond contributing -6.1 units.
The total change at SOooK is 6so
t
= -3.S or log A = lZ.8.
3.3 RECOMBINATION OF ALPHA RADICALS
This is the reverse of reaction (D5),
Ph-CH
Z
' + Ph'-CH
Z
-+ Ph-CHZCHZPh' (R1)
As mentioned earlier,reactions (D5) and (R1) are diffusion limited when they take
place in the condensed coal phase. The A-factor estimated below assumes diffusion
free recombination, therefore is strictly limited to gas phase recombination.
Under gas phase conditions the activation energy of (R1) is zero. The activa-
tion entropy is related to the activation entropy of the reverse reaction as follows
6so
t
(R1) = 6so
t
(D5) + 6S0(R1)
where 6S0(R1) is the entropy of the reaction. 6S0(R1) can be estimated by group
additivity. At SOOOK we have
6So (R1) = Z[C-(CS)(C)(H)Z]e - Z[C.-(Cs)(H)Z]e = -35.3
where the subscript e denotes entropy. vJe also have from section 3.1, 6so
t
(D5) 1.1
so that 6so
t
(R1) = 1.1 - 35.3 = -34.Z. For a bimolecular reaction
k
s
T RT 2 t
A =----- ---- e exp(6S0 /P)
h P
o
30
with the state being at pressure Po= 1 at. For the entropy change of
-34.Z, log A = 8.4 /s g-mol.
3.4 HYDROGEN ABSTRACTION
The following are a few representative reactions
X' + Ph-CH
3
+ XH + Ph-CH
Z
'
X = H
X= CH
3
X = C
Z
H
5
X' + Ph-CH
Z
CH
3
+ XH + Ph-CHZCH
Z
'
X H
X CH
3
X C
Z
H
5
(HAl)
(HAZ)
(HA3)
(HA4 )
(HA5)
(HA6)
Ph-CHZCH
Z
' + + Ph-CH
Z
CH
3
+ (HA7)
The activation energies of (HAl), (HAZ) have been estimated as Z.3, 8 by a
variant of the bond-order-bond-energy method (ref. 40). The activation energy
of (HA3) can be estimated by analogy to suitable aliphatic analogs as follows.
The two reactions
CH
3
' + (CH
3
)3CH + CH
4
+ (CH
3
)3C,
C
Z
H
5
' + (CH
3
)3CH + C
Z
H
6
+ (CH
3
)3C'
have activation energies 8 and 8.9 respectively. The difference of 0.9 is assumed
to apply to the reactions (HAZ) and (HAl) as well, producing the estimate 8.9
for (HA2).
The activation energies of reactions (HA4)-(HA6) are taken equal to those of
their aliphatic analogs. For example, the aliphatic analog of (HA5) is
CH
3
' + CH
3
CH
Z
CH
3
+ CH
3
CH
Z
CH
Z
' + CH
4
Finally, the activation energy of (HA7) is taken equal to that of (HA3).
The estimation of A-factors for hydrogen abstraction reactions is illustrated
by the following two examples.
Example 3.13
The transition relevant to (HAl) is
CM-Ph-CH
3
+ H' + CM-Ph-CHZH H (hal)
By definition, the entropy of this transition is
6so
t
= sot-SO(R)-SO(H.)
where R represents CM-Ph-CH
3
. The difference sot-SO(R) has the following components
(i) Translation, external symmetry and rotation zero or negligible.
(ii) Spin from Z to 1 with contribution RnZ.
(iii) The barrier for internal rotation about the Ph-C bond changes from 0 to 7.5
kcal contributing -1.3 units. At the same time a loss of a symmetry factor of 3
31
associated with internal rotation contributes R ~ n 3 units,
(iv) A C-H stretch becomes reaction coordinate causing an entropy change of -0.1.
A new HH stretch (2800 cm-
l
) contributes 0.1.
(v) Two new HHC bends (1000 cm-
l
) contribute 2 x 1.1 = 2.Z.
(vi) Two H-C-H Bends (1400 cm-
l
) become H "C-H bends (1000 cm-
l
) causing an
entropy change of Z x 0.5 = 1.
The resultant of these components is sot_SO(R) = 5.5. At the same time standard
tables (ref. 30) give SO(H') = 3Z.3 so that 6s
ot
= -Z6.8 and log A= 10.0.
The A-factors of reactions HAZ and HA3 are best estimated by comparison with
aliphatic analog reactions
H' + C
Z
H
6
+ HZ + C
Z
H
5
' (HAl')
CH
3
' + C
Z
H
6
+ CH
4
+ C
Z
H
5
' (HAZ')
C
Z
H
5
' + C
Z
H
5
COC
Z
H
5
+ C
Z
H
6
+ C
Z
H
5
COC
2
H
4
' (HA3')
The experimental log A for these reactions are 11.0, 8.5, 8.0 at 400
0
K (ref. 30).
Assuming that log A (HAi)-log A (HAl)=log A (HAi')-log A (HAl') for i = 1,Z,3
and that each difference is independent of temperature we obtain for 800
0
Kthe
crude estimates log A (HAZ) = 7.5, log A (HA3) = 7.0.
Example 3.14
For reaction (HA4) we need the entropy change of
H + CM-Ph-CH
Z
CH
3
+ CM-Ph-CHZCHZH H (ha4)
This change can be calculated by analogy to (hal) from which it differs only by
the fact that the barrier for internal rotation about the Ph-C bond does not
change. Therefore, we set 6s
ot
(HA4) = 6s
ot
(HAl) + 1.3 = Z5.5 and log A (HA4) = 10.3.
Reactions (HA5) and (HA6) can be treated similarly to (HAZ) and (HA3). The
result is log A (HA5) = 7.8, log A (HA6) = 7.3. Finally, in the absence of any
other information we can crudely set log A (HA7) = log A (HA6).
energies of (AI) and (AZ) can be approxi-
3.5 ADDITION OF RADICALS TO DOUBLE BONDS
These reactions are the reverse of (DBl)-(DB4) and may be exemplified
X' + Ph-CH CH
Z
+ Ph-CHCHZX
X' + Ph-CH = CH-Ph' + Ph-CHCH-Ph'
I
X
where X= H,CH
3
or C
Z
H
5
. The activation
mated by those of the aliphatic analogs
X' + CH
3
CH
Z
CH = CH
Z
+ C H 3 C ~ Z C H C H Z X
X' + CH
3
CH = CHCH
3
+ CH
3y
HCHCH
3
X
by;
(ADl)
(ADZ)
These are listed in the compendium of Kerr and Parsonage (ref. 41) as 1.Z,7.Z,
7.3 for (ADl') with X= H,CH
3
, C
Z
H
5
, respectively. The corresponding values for
(ADZ') are listed as 4.3,7.5,8.5.
32
The e n t r o ~ i e s of activation of (ADl) and (AD2) can be calculated from the
entropy of reaction and the activation entropy of the reverse reaction. Taking
as an example (ADl) with X= CH
3
we have
Llsot(ADl) = Llso
t
(DB2) + LlSO(ADl)
Using group additivity we calculate for 800
0
KLlSo= -33.7 so that Llsot(ADl)= 3.4-
33.7 = -30.3, log A = 9.3.
3.6 ADDITION OF RADICALS TO AROMATIC RINGS
The limited literature on the addition of small radicals like CH
3
' and C
2
H
5
'
to aromatic rings has been reviewed by Szwarc et al. (refs. 42,43) and Williams
(ref. 44). The reaction mechanism consists of addition to the aromatic ring to
form a cyclohexadiene intermediate followed by dissociation as shown in the follow-
ing two examples:
3 .. :0+ CHi
(a)
Addition reactions have been studied in solution wherein methyl and other car-
bon-centered radicals were produced by the decomposition of initiators like acetyl
peroxide. Hydrogen addition has not been studied as such because it cannot be
produced in solution in any clean-cut fashion. However, hydrogen addition at temp-
eratures higher than 400C has been identified as an important step in the thermal
reactions of various aromatic compounds (ref. 33,45).
When addition takes place in solution at modest temperatures, the intermediate
cyclodiene radical can recombine with other radicals or abstract hydrogen leading
to a variety of products. At temperatures representative of coal pyrolysis (above
400C) the decomposition steps 2 or 3 in (a) or 5,6 in (b) would be quite rapid
to effectively suppress recombination reactions. Because of the larger energy of
H' compared to CH
3
' the decomposition step 3 is much faster than (2) and reaction
33
(a) is kinetically equivalent to
Ph-CH
3
+ H" + PhH + CH
3
"
Likewise,as a result of the higher stability of the benzyl radical
+ CH
3
" + Ph-CH
3
+ Ph'-CH
2
' (b')
Because reactions (a) or (b) often follow the route (a') or they are some-
times characterized by the terms aromatic substitution or displacement reactions.
Returning to reaction (a) we note that the addition of the hydrogen atom can
take place at other positions, for example
+ H
(a")
This reaction is obviously kinetically insignificant. Inasmuch as reported ex-
perimental values for the rate constants of addition reactions refer to addition
at any position around the ring, these constants should be suitably reduced if
addition at a specific site is considered. Unfortunately, the relative addition
rates at positions like (a) and have not yet been measured.
In coal pyrolysis, reactions of the type (a) can be the source of various hy-
drocarbon gases via the corresponding radicals. For example the CH
3
' radical
produced in (a) will evolve as methane after hydrogen abstraction from some suit-
able site. Reaction (b) is important as a low activation energy route to the
dissociation of methylene bridges.
The rate parameters in Table 3.2 refer to addition to the benzene ring
y
U + x
y
(}x
(Ai)
where X= H,CH
3
,C
2
H
S
(i=1,2,3) and Y is Hor an alkyl chain or bridge. Depending
on the relative stability of X' and y., the decomposition of the cyclohexadiene
radical would lead to the product PhX or back to the reactant PhY. At high temp-
eratures the dissociation of the hexadiene radical is relatively rapid so that
the overall reaction is controlled by the rate of the addition step. The activa-
tion energies of the addition reactions are in the range 0-8 (ref.30), probably
close to 8 for CH
3
' and C
2
H
S
' and about 2-4 for H'. parameter values listed
in Table 3.2 refer to the benzene ring. It must be noted that the rates of addition
increase with increasing size and decreasing stability of the aromatic ring system.
Szwarc and Binks (ref.43) report relative rates of 1,22,27 and 280 for the addition
of methyl radical to benzene, naphthalene, phenanthrene and anthracene, respect-
34
ively. The rate constants must obviously be adjusted upwards when applied to
kinetic modeling of coal reactions.
3.7 REACTIONS OF CARBOXYL AND PHENOLIC HYDROXYL GROUPS
Among the principal products of pyrolysis, water and carbon dioxide are gener-
ally attributed to the phenolic hydroxyl and carboxyl groups in coal. Brooks
et al. (ref.46) used IR spectroscopy and wet methods to the con-
centration of oxygen groups in brown coals undergoing pyrolysis and suggested
that at temperatures below 300C water is produced by esterification reactions
presumably of the type
Ph-COOH + Ph'-OH Ph-COO-Ph' t H
2
0 (i)
At temperatures between 300 450C they observed a further decrease of phenolic
-OH by reactions of the type
( i i )
Ph-OH + Ph'-CH
3
Ph-CH
2
-Ph' + H
2
0
Ph-OH + Ph'-OH Ph-O-Ph' + H
2
0 (iii)
of which (iii) was singled out as more important. At the same time, ester groups
formed by (i) were found to decompose with the elimination of carbon dioxide,
Ph-COO-' Ph-Ph' + CO
2
(iv)
Presumably, free carboxyl groups would also decompose under these conditions.
Reactions (i)-(iii) are significant not only as sources of "chemical water"
but as producing additional linkages among the structural units resulting in a
suppression of tar formation. Since the content of hydroxyl and carboxyl groups
is higher in coals of low rank, the above condensation reactions offer an explana-
tion for the low amount of tar produced in the pyrolysis of subbituminous coals
and lignites. The kinetics of the condensation reactions (i)-(iii) have not been
studied at the temperatures of interest to pyrolysis (above 400C). Moreover,
the experimental evidence has not as yet been sufficient to distinguish which of
these three reactions is more important.
At temperatures above 600C the reactions of hydroxyl groups become more complex
leading to a variety of products including carbon monoxide. While all reactions
discussed in sections 3.1-3.6 were free radical in nature, the high temperature re-
actions involving phenolic hydroxyls are believed to largely proceed by concerted
mechanisms. The remainder of the chapter is a survey of concerted mechanisms
relative to the thermal reactions of coal.
Concerted Reactions
The term "concerted" indicates reactions that involve simultaneous breaking
(and making) of more than one bond. By contrast, the free radical reactions sur-
veyed so far involve the breaking (and making) of one bond at a time. The follow-
ing examples drawn from model compound studies give an idea of the role of con-
certed reactions in coal pyrolysis.
35
Cypres and Betten (refs. 47-49) studied the pyrolysis of phenol, o-cresol and
p-cresol labelled at specific positions by 14C and 3H. The reactions were carried
out in the temperature range 700-900C and resulted in a wide variety of products.
For example, in the pyrolysis of one mole of o-cresol at 750C for 2.5 seconds,
the major products (in moles) were 0.064 benzene, 0.034 toluene, 0.242 phenol,
0.10 water, 0.139 carbon monoxide, 0.128 methane, 0.078 hydrogen and 0.032 char.
When the o-cresol was tritiated on the hydroxyl group, the toluene contained a
considerable amount of tritium while the water contained 0.61 atoms of tritium.
This distribution of tritium was interpreted as evidence against the free radical
mechanism (i) which would produce toluene free of tritium. Instead, the authors
proposed the concerted mechanism (ii) which produces tritiated toluene. This
explanation is not complete, however, since the elimination of oxygen in the last
step of (ii) is not mechanistically satisfactory as written down and,in addition,
it does not explain the formation of a considerable amount of tritiated water.
Mechanism (i) can partially explain the formation of tritiated toluene and water
if H is replaced by H* as shown in (i)
OH*
*HO
H
(yCH. (yCH.
UCH.
I +
H
--.
I --. I +
OH* (i)
~ ~
~
OH*
*H 0
(yCH.
HVCH.
* CH
3
2H
((+
~ I
...
I
OH
( jj)
~
OH*
*H(:iCH. (yCH.
(yCH.
~ I
+ H*
...
I ------.
~ I +
OH*
( i I)
~
~
The relevant issue in choosing between (ii) and (i) is the relative rates of the
two reactions producing hydrogen atoms:
36
OH*
OH*
(y
CH
2
+ H ( iii)
(yCH, I
~
~ I
O
(yCH, +
H* ( iii I )
~ I
Mechanism (ii) provides a better explanation of the products when (iii) is faster
than (iii ') while (i') becomes more credible when (iii') is faster. Unfortunately
the relative rates of (iii) and (iii') have not been determined independently.
Another possible explanation is hydrogen exchange (scrambling) whereby the tritium
atom is spread around the ring preceding dissociation. It appears that the ex-
perimental data available are not sufficient to discriminate between the free
radical and the concerted mechanisms.
To explain the formation of carbon monoxide the authors of refs. 47-49 suggested
two possible schemes, one of which involves a seven-membered ring,
OH
OH
0
H +

H +
6
-O+co
CH2
Independently of the validity of this particular mechanism, it is clear that the
formation of carbon monoxide requires the disruption of the benzene ring and cannot
be explained solely by free radical steps. It should be noted in this connection
that carbon monoxide appears in significant amounts only when the pyrolysis temp-
erature exceeds 700C.
In an attempt to demonstrate the importance of pericyclic pathways in the
thermal reactions of coal, Virk et al. (ref.50) studied the hydrogenation of
anthracene, phenanthrene and a high volatile bituminous coal with various hydrogen
donor solvents at 300C for two hours. One of the reactions studied was
37
The percentage conversion obtained with different H-donor solvents was as follows;
CD
o
O
3 5 58
The large difference with the last two solvents was explained by the Woodward-
Hoffman rules governing concerted reactions.
It must be noted that if reaction (iv) were to take place by a free radical
mechanism it would require as a first step the dissociation of a hydrogen atom
from the H-donor molecule;
+ H
( i V ')
This reaction has activation energy about 69 kcal, therefore would be extremely
slow at 300C. Using stereospecificity properties, von E. Doering and Rosenthal
(ref. 51) demonstrated a concerted path for the thermal decomposition of cis-9,
10-dihydronaphthalene to naphthalene, tetralin and hydrogen.
Several authors who have studied the pyrolysis of tetralin or the dissociation
of bibenzyl in tetralin (refs. 52-54) have explained their results solely by free
radical mechanisms. More experimental work is needed, preferably with labelled
compounds, to determine which of the pyrolysis reactions proceed by concerted
mechanisms and which by the more widely accepted free radical mechanisms.
TABLE 3.2 Rate parameters of selected elementary reactions at 800
0
K
loglOA
E
(A in
-1
s) (kcal/g-mol) s or 1t/g-mol
D1 14.9 BN 85.3 G
D2 15.3 BN 72.4 G
D3 15.4 BN 69.5 G
D4 14.3 G 80.7 G
D5 13.9 BN 56.4 G
D6 15.3 G 68.6 G
D7 14.3 G 85.5 G
D8 13.9 G 55.6 G
38
OBI 15.1 G 51. 7
G
DB2 14.4 G 45.0 G
DB3 14.2 G 9.6 G
DB4 12.8 G
50.2
G
Rxn
lO9l0A
E
Rl 8.4 a
HAl 10.0 G 2.3 ZM
HA2 7.5 G 8.0 ZM
HA3 7.0 G 8.9 G
HA4 10.3 G 9.7 BTK
HA5 7.8 G 10.8 BTK
HA6 7.3 G 13.4 ZM
HA7 7.0 G 8.9 G
ADI
X=l 11.0 G 1.2 G
2 9.3 G 7.2 G
3 9.2 G 7.3 G
AD2
X=l 4.3 G
2 7.5 G
3 8.5 G
Al
X=l 9.3 G 1.2 KP
2 7.3 G 7.2 KP
3 7.3 G 7.3 KP
BN: ref. 29
BTK: ref. 30
G: estima ted by the author
KP: ref. 41
ZM: ref. 40
39
Chapter 4
EXPERIMENTAL TECHNIQUES AND RESULTS IN FLASH PYROLYSIS
4.1 EXPERIMENTAL TECHNIQUES
As set out in the general introduction, the survey of experimental results will
be confined to flash pyrolysis at the exclusion of slow pyrolysis or carbonization.
Flash pyrolysis poses three experimental difficulties that need careful considera-
tion: (i) control and measurement of the temperature-time history of the coal par-
ticles (ii) suppression of secondary reactions (iii) quantitative collection of
products.
The temperature-time history of the coal particles generally consists of a
heating period, a period at approximately constant temperature and a cooling or
quenching period. While isothermal operation permits the easiest kinetic analysis
of the results, the reaction occurring during the heating and cooling times is often
significant. The kinetic analysis of the data in this case must take into account
the full temperature-time history. Whether or not the pyrolysis occurs isotherm-
ally, the measurement of the coal particles' temperature is not trivial. In many
cases the temperature cannot be directly measured but must be calculated from a
heat transfer model. The other two experimental problems,the suppression of
secondary reactions and the collection of products, depend on the reactor
geometry and flow pattern and are best discussed separately for the entrained
flow and the captive sample techniques.
4.1.1. The entrained flow technique
In this experimental set-up shown schematically in Fig. 4.1, coal particles
20-100 ~ m are carried by a primary stream of an inert gas through a water-cooled
injector at the axis of a vertical furnace. A secondary and larger stream of
inert carrier gas flows downward alonQ the furnace under laminar flow to assure
that the particles are not dispersed radially to the furnace walls. Particles
and volatile products are collected by a water-cooled probe of special design
ensuring representative samples. The reaction time is adjusted by the flowrate
of the secondary stream and the axial location of the probe. Heating elements
around the furnace produce a uniform wall temperature in the middle zone of the
furnace. The secondary gas stream is pre-heated to decrease the heating time.
A system of this type has been used by several workers (refs. 55-59). Kobayashi
et al. (ref. 57) discuss in detail the experimental aspects of the technique and
provide an analysis of the heating history of the particles including the effect
of injector geometry and the mixing between primary and secondary gas streams.
Their calculations indicate heating times in the range 5-50 ms. The times required
for 40% weight loss of a bituminous coal varied in the range 10-100 ms as the
40
furnace ranged from 2,100 to 1,500
o
K. Since the pyrolysis could not
be assumed to occur at constant temperature, the kinetic analysis of the data in-
corporated suitable heat transfer calculations.
Pre heater
r
Secondary Stream
Furnace
t Primary Stream
(cool + carrier gas)
Sampling Probe
To Gas Ana lysis

Fig. 4.1. Schematic of an entrained flow pyrolysis furnace. (source: ref. 59).
A second area of concern in the entrained flow technique is the secondary re-
actions suffered by the volatile reaction products during their residence in the
furnace. These reactions make the mechanistic interpretation of product distrib-
ution somewhat doubtful but are of no great concern in combustion studies where
the primary information required is weight loss and perhaps elemental composition
of the volatiles.
The third experimental aspect of the entrained flow technique that requires
careful consideration is the collection and analysis of products. In the afore-
mentioned investigation (ref. 57) the weight loss was determined both directly
and indirectly. The direct technique consited of simply weighing the char collected
in the probe. It was independently shown by cold flow experiments that the col-
lection efficiency of char particles was 95 to 98%, suggesting similar accuracy
in the measurements of the weight loss. The indirect technique employed the coal
ash as a tracer. The results from this technique are independent of collection
efficiency but are subject to some error due to vaporization of mineral matter
41
components.
In addition to the weight loss, certain of the gases (CO,COZ,SOZ,NOx) were de-
termined by on-line continuous analyzers (ref. 59) or by gas chromatography (ref.
57). In some studies the solid char was characterized by Fourier-transform
infrared spectroscopy (ref. 59). However, products of molecular weight above ZOO,
usually classified as tar, could not be collected quantitatively. Condensation on
the probe, the filter, the char, the walls of the gas collection vessel or the
walls of the furnace itself preclude a quantitative analysis.
The furnace, injector, probe etc. described in references 57 and 59 represent
the most careful and up-to-date designs. Earlier investigations using the en-
trained flow technique include that of Badzioch and Hawksley (ref. 56) who also
used laminar flow conditions and that of Eddinger et al. (ref.55) used both
laminar and turbulent flow conditions. The last emplcyed a large coal-to-
gas mass flow ratio to provide sizeable tar The increaspd amount of
solids, however, interferred with the temperature measurements and increased the
extent of secondary reactions.
4.1.Z The captive sample technique
A good description of this technique can be found in Anthony et al. (ref. 60).
A small sample (5-Z00 mg) of ground coal is placed between the folds of a wire-
cloth screen (see Fig. 4.2a) heated by a DC or AC current I(t). The resistive
assembly is attached inside a metal shell containing an inert gas or hydrogen.
The gas pressure-can be varied from vacuum to 100 at. In addition to providing
the inert atmosphere at the desired pressure, the metal shell serves as a product
collection vessel.
The coal sample is placed at the center of the screen where the temperature
profile is relatively uniform. Near the electrodes the temperature drops due to
conductive losses. A fast response (low mass) thermocouple serves to measure
the temperature-time history. The sample size is chosen as a compromise of two
considerations. A large sample minimizes the effect of coal inhomogeneity and
generates a sufficient quantity of products for analysis. At the same time,
however, a large sample of particles creates some irreproducibility because of
possible rearrangement of the particles on the screen during heating. Samples
of about 200-500 mg have been used by Solomon and Colket (ref.61) and Gavalas
and Wilks (ref.62) allowing heating rates as high as 500C/s and final tempera-
tures as high as lZOOC. In the experiments of Anthony et al. (ref.60) the sample
size was 5-10 mg producing heating rates up to 10,000C/s with final temperatures
as high as lZOOC.
T
0)
T
T,
Fig. 4.2. The captive sample technique: (a) pyrolysis apparatus (b) single-
pulse T-t response (c) double-pulse T-t response.
The temperature-time history T(t) experienced by the coal sample depends on
the current input I(t), the geometry of the resistive assembly and surrounding
shell and the nature and pressure of the surrounding gas. For given sample size
and gas pressure, and for constant current I, the sample temperature reaches a
steady state Ts=f(I) at which the resistive heat input balances losses by con-
duction, convention and radiation (Fig. 4.2b). The time required to reach the
steady temperature is usually on the order of seconds; therefore, constant
current inputs are not suitable for producing large heating rates. To circumvent
this problem, Anthony et al. (ref.60) applied current inputs consisting of two
43
consecutive pulses of constant current. If I
Z
is the current required to produce
the steady temperature T
Z
' then switching from II to I
Z
approximately when the
temperature has reached T
z
, produces the response shown schematically in Fig. 4.Zc.
The heating rate is thus limited only by the maximum current II of the power supply.
Upon switching from II to I
Z
' the temperature overshoots slightly before settling
to a constant level TZ(Fig. 4.Zc). The overshoot can be minimized by a proper
choice of the current II and the switching time. After being held at T
Z
for the
desired time, the sample is cooled by turning the current off. While the cooling
period may last for a couple of seconds, after the first ZOO ms or so the temperature
becomes too low for the pyrolysis reactions to continue.
The most important items in product analysis are weight loss (total volatiles)
and tar yield. The weight loss is determined by weighing the coal sample before
and after pyrolysis. The tar condenses on aluminum foils lining the walls of
the vessel. These foils can be removed and weighed after the completion of an
experiment. The tar can be subsequently dissolved in a suitable solvent, e.g.
tetrahydrofuran, for elemental analysis or nmr spectroscopy. Because of the short
duration of each experiment the pyrolysis vessel remains cold (T <50C in most
cases) and the recovery of tar on the aluminum foils is quite high. A small amount
of tar remaining in the gas volume as vapors or aerosol can be collected by
flowing the vessel contents through a filter (refs.63,64).
The gaseous products can be analyzed directly from the vessel or after collect-
ing in suitable cold traps. Direct collection was used with sample sizes of 100-
ZOO mg producing measurable concentrations of gases in the vessel (refs.61,6Z,65).
Some experiments (ref.65) indicated that a nonnegligible fraction of the hydro-
carbon gases were retained in solution in the tar deposited on the aluminum foil
or on various surfaces in the vessel. To recover the gases quantitatively, it
was found necessary after each pyrolysis run to heat the vessel to about 150-Z00C
and thus drive the hydrocarbIDn gases to the gas phase. The heating in this case
was accomplished by a mobile oven sliding on rails (ref.65). A disadvantage of
this technique is that at 150-Z00C a certain fraction of light products in the
tar would also enter the gaseous phase but would not be detected by the gas chroma-
tographic procedure employed. The indirect technique of gas collection was de-
veloped by Suuberg (refs. 63,64) to concentrate the products derived from small
sample sizes (about 10 mg). The contents of the vessel were flown through two
cold traps packed with Porapak Q. The first trap maintained at room temperature
retained intermediate products such as benzene, toluene and xylene. The second,
maintained at -196C by liquid nitrogen retained all gases except HZ. The products
were recovered by warming the traps at Z40 and 100C respectively.
In both the direct and the indirect technique the gaseous products were anal-
yzed by gas chromatography employing a thermal conductivity detector for CO, CO
Z
'
44
H
2
, H
2
0 and a flame ionization detector for hydrocarbons. The determination of
water caused persistent difficulties. One difficulty was the uncontrolled adsorp-
tion-desorption of water from pyrolysis or from the atmosphere on vessel walls
and gas lines. The other difficulty involved the integration of the broad and
distorted chromatographic peak. Solomon and Colket (ref. 61) used a different
and somewhat more reliable procedure for water determination.
The overall efficiency of product collection has varied among the various
experimental setups employed. The best results seem to have been obtained by
Suuberg, who reported (refs. 63,64) a 95% total mass balance and 90% balance on
carbon and hydrogen. The largest analysis error is in water, as much as 40%, the
lowest in hydrocarbon gases, 5-10%.
An interesting version of the captive sample technique has been described by
JUntgen and van Heek (ref. 66). The main features of the experimental setup was
direct connection of the pyrolysis vessel with a time-of-flight mass spectrometer
permitting direct on-line analysis of gaseous pyrolysis products and a facility
for movie-camera recording of physical changes of the coal particles. A controlled
power supply provided a linear temperature-time profile with heating rates up to
3,000C/s.
4.1 .3 The" pyroprobe "
In a variation of the captive sample technique, the coal sample is placed in a
heated probe, the "pyroprobe", directly connected to the injection port of a gas
chromatograph. The commercially available pyroprobe consists of a platinum ribbon
as the heating element with associated power supply and control circuitry. Using
coal samples .of less than 5 mg, the heating element can supply heating rates up
to 20,000C/s and final temperatures up to 2,000C. A helium or other inert
carrier sweeps the pyrolysis products through a short line into the chromatographic
column.
Among the reaction products, gases and compounds of intermediate volatility can
be analyzed by using suitable separation columns. For example, capillary columns
permit the elution of compounds as heavy as naphthalene and phenol. Heavy tars
condense on tube walls or column packing and cannot be analyzed. The char residue
is weighed at the conclusion of each experiment. Applications of the pyroprobe
have been reported in refs. 67 and 68.
To conclude this section we summarize the relative merits of the three tech-
niques described. The entrained flow technique is suitable for high temperature
and short residence time pyrolysis where it provides the best temperature control
and rapid quenching. The steady state operation allows the processing of a large
quantity of coal to smooth sample inhomogeneity. Although gas and char collection
is straightforward, tar collection is difficult. The captive sample technique
involves much simpler apparatus and allows arbitrary pressure (including vacuum) and
45
residence time. Except at the highest temperatures, it allows good control of
temperature and heating rates. However, high heating rates can be achieved only
with small samples (- 10 mg) exacerbating the problem of sample inhomogeneity.
Product collection is good, although the milligram quantity tar collected in the
case of small coal samples is insufficient for chemical characterization. Com-
pared to the standard version of the captive sample technique, the pyroprobe
arrangement is limited by small sample size and operation at close to atmospheric
pressure. In addition, the pyroprobe does not allow the collection of heavy
products. However, the direct injection of products into the chromatographic
column is a very convenient feature and greatly reduces the turnaround time for
an experiment. A more detailed discussion of the operating characteristics of
various experimental arrangements is given in a recent comprehensive report by
Howard et al. (ref. 69).
4.2 EXPERIMENTAL RESULTS AND DISCUSSION
This section contains a survey of data on weight loss, product distribution
and product composition as functions of temperature and pyrolysis time. The
bulk of the data reported here derive from essentially isothermal experiments,
small particle size and low pressure (vacuum to 1 at). Limited results involving
variations in the heating rate and temperature-time history will be discussed in
conjunction with the kinetic modeling in chapter 6. Pressure and particle size
are variables which affect the rate of transport processes, therefore they will
be discussed i ~ chapter 5 on heat and mass transfer. The results discussed in
the present chapter relate to conditions which mimimize interferences by trans-
port phenomena and the concommitant secondary reactions. Our survey of experi-
mental data is selective rather than extensive, emphasizing recent comprehensive
work and altogether omitting earlier or more narrowly focussed studies. For a
broader experimental survey the report of Howard et al. (ref. 69) is highly rec-
ommended.
4.2.1 \ ~ e i g h t loss
Although only an overall measure of the reaction's progress, the weight loss
or total yield of volatiles displays a complex temperature dependence. To facil-
itate the discussion we introduce a few terms:
cumulative yield:
instantaneous yield:
fraction or percentage of the weight of a product
evolved during the period of pyrolysis relative to
the weight of coal on an "as received" or dry ash-
free basis; the cumulative yield is generally a func-
tion of temperature and time.
the derivative with respect to time of the cumulative
yield.
46
yield:
ultimate yield:
the term yield will be used as a short-cut for
cumulative yield.
at fixed temperature the cumulative yield asymp-
totically reaches a maximum value, within exper-
mentally reasonable times; this maximum value will
be called the ultimate yield and is generally a
function of temperature.
B
The ultimate yield defined above has relative rather than absolute significance.
Consider for example two first order parallel and independent reactions with the
same product,
k
1
A - - - ~ P
k
2
- - - ~ P
If A
o
' B
o
are the initial amounts of A,B, the yield is
A B
~ (l_e-kl
t
) + ~ (1_e-
k
2
t
)
W W
and the theoretical ultimate yield is (Ao+Bo)/W. Suppose, however, that k
2
has
a much higher activation energy than k
1
such that at low temperatures k
2
k
1
.
Then if the pyrolysis time is limited for practical reasons, the ultimate yield
at low temperatures will be Ao/W. At sufficient high temperatures, however, the
constant k
2
will be significant and the ultimate yield will reach its theoretical
value. Since pyrolysis involves a large number of reaction steps with a wide
range of activation energies,the (apparent) ultimate weight loss is an increasing
function of temperature. In principle,the ultimate weight loss also depends on
the temperature-time history because of the coupled nature of the chemical reactions.
However, the dependence on the temperature-time history is rather weak as will be
discussed in Chapter 6.
Representative weight loss data from an entrained flow system and a captive
sample system are shown in Fig. 4.3 and Figs. 4.4, 4.5 respectively. Figure 4.3
shows the weight loss obtained by the entrained flow technique at high temperatures
and short residence times (ref. 57). The data points represent cumulative weight
loss corresponding to the pyrolysis time under question. Each point, therefore,
corresponds to a distinct experiment employing a different sample of coal particles.
The scatter in the data, as high as 20%, is due to measurement error as well as
sample variation. Another noteworthy feature is the inflection point at the early
stages of pyrolysis, indicating an initial acceleration of the reaction rate.
Since the data have been corrected for the effect of heating time, this initial
acceleration which has also been observed in other studies seems to be related
either to the presence of consecutive reactions or to intraparticle mass trans-
fer retardation.
47
To prevent secondary reactions o ~ t ~ r vaoors on particle surfaces influencing
the weight loss, it is necessary to k e e ~ very low rarticle density in the en-
trained flow reactor. Secondary reactions on the reactor walls would still take
place but would only affect the product distribution and not the weight loss.
80,..----------------------,
Lignite
LL

o
Bituminous Cool
~ 80.----------------------,
o
.r::.
.? 60
Q)
~
20 40 60 80 150 200
Reoction time (ms)
Fig. 4.3. Weight loss vs. pyrolysis time at various furnace
temperatures (source: ref. 57).
Solomon and collaborators (refs. 61,70) have studied the pyrolysis of a large
number of coals under vacuum (20-60 mm Hg) by the captive sample technique. Fig-
ures 4.4,4.5 are representative samples of Solomon's results showing the total
weight loss as a function of pyrolysis temperature for two different residence
times, 20s and 80s. The weight loss (cumulative) has considerable scatter, some
of which may be due to sample variability but a major fraction is probably due
to fluctuations in the sample temperature and loss of fine coal fragments from the
screen. The solid curves were calculated by a kinetic model discussed in Chapter 6.
48

80
u..

o
-oe 60
20
t:. 205
o 80S
o
Pyrolysis temperature (OC)
Fig. 4.4. Weight loss vs. temperature for a hvc bituminous
coal "Ohio No.2" (source: ref. 70).
100
{; 205
0 80S
80
u.. 0 0

0
<t
60

5!

40
0-

20 {;
0
300 500 700 900 1100 1300
Pyrolysis temperature
IOe I
Fig. 4.5. Weight loss vs. temperature for a hva bituminous
coal "Lower Kitanning" (source: ref. 70).
49
The results of Figure 4.4 exhibit some important trends. Above about 700C
the weight loss at 20 and 80s is the same, within experimental error, indicating
that 20s is adequate to attain the ultimate asymptotic value. At temperatures
below 700C the deviation between the results at the two residence times is sig-
nificant indicating that 20s is insufficient for achieving the ultimate weight
loss.
Figure 4.5 presents the weight loss for a bituminous coal with the rank "high
volatile A". The main difference with Figure 4.5 is in the ultimate weight loss
which seems to change very little with temperature above 900C. This difference
will be explained below in the discussion of individual product yields.
4.2.2 Product distribution
The product distribution is the most essential Information relative to tne
commercial utilization of pyrolysis and at the same time sheds considerable light
on reaction mechanisms. Representative experimental data are shown in Figs. 4.6-
4.8 taken from Solomon's work (ref. 70). They present the cumulative yields of
tar, H
2
0,C0
2
,CO,H
2
and hydrocarbon gases at residence times of twenty seconds.
The balance is the residual solid, char. The label T + Mdenotes tar and "missing"
material that escaped the collection procedure and is estimated only by an overall
mass balance. This missing material probably consists largely of tar whence
lumped with the collected tar. The solid lines again represent results of model
fitting. Despite the cumulative nature of the product yields, the scatter in the
data is considerable demonstrating the difficulties inherent in such measurements.
Measuring the instantaneous (or differential) yields in this experimental setup
is cl early impracti ca1.
The products can be classified into two groups relative to the temperature de-
pendence of the yields. Tar, water and carbon dioxide evolve at lower temperatures
with ultimate yields that are essentially independent of temperature above 700C.
The second group of products consisting of gaseous hydrocarbons, carbon monoxide
and hydrogen evolve at higher temperatures. The ultimate yield of these products
continues increasing with temperature up to 1,OOOC or higher.
Coal rank is a very important factor in the distribution and temperature de-
pendence of various products. In bituminous coals, tar makes up 50-80% of the
weight loss, the remaining consisting of hydrocarbon gases, water and carbon oxides.
In subbituminous coals,water and carbon oxides are produced at increased yields,
as much as 60% of the weight loss, while tar contributes only 25-40%. In lignites
tar is even lower and gases higher as illustrated in Fig. 4.6.
50
5.0
+
+
+
+
4.0
"
"-
o CO
2
Water
<l
+
Waler
0
o Tor. M/20
;/.
o HCl3
+
li. eO/5
.c 3.0
"V Hydrogen
CO,
O'
.0;
3
:2
OHC/3
"
2.0 .>'
u
V

Hydrogen
"0
0
0
a:
10
TOr + M/20
0
0
300 1100 1300
Fig. 4.6. Product yields vs. temperature for a Montana lignite
at 20s pyrolysis time (source: ref. 70).

Water
;: 4.0
<l
o
3.0
.0;
3
"0
";., 2.0
u

"0
o
a:
10
o CO
2
+ Waler
o Tor+M/20
o HC/3
t::. eO/5
V Hydrogen
+
+
+
CO/2

Fig. 4.7. Product yields vs. temperature for a hvc bituminous
coal "Ohio No.2" at 20s pyrolysis time (source: ref. 70).
51
The variation of the relative distribution of tar and gases among coals of
different rank explains the previously observed temperature dependence of the
ultimate yield. In bituminous coals where the weight loss is dominated by tar,
the ultimate yield appears to increase very little beyond 700
0
e. In subbituminous
coals and lignites, where a considerable fraction of the volatiles consists of eo
and hydrocarbon gases, the ultimate weight loss continues increasing with temper-
ature even beyond 1,0000e.
Tar is the most abundant and commercially valuable product from the pyrolysis
of bituminous coals. It is a mixture of many compounds with molecular weights
mainly in the range 200-1200. At the temperature of pyrolysis it is produced as
a vapor but at room temperature it becomes a viscous liquid or solid. The tar
liquids consist largely of dimers of smaller fragments generated by the primary
bond dissociation reactions.
In an ideal experimental setup, once released from the coal particles the tar
molecules are removed from the high temperature region escaping secondary reactions.
The captive sample and entrained flow techniques approximate this desirable oper-
ation. Figure 4.9 shows the cumulative tar yield at two pyrolysis times, 20s and
80s, as a function of temperature. Despite the large scatter in the data, the
ultimate yield (80s) is clearly independent of temperature above 500
0
e. As a
matter of fact, the yield at 80s seems to slightly decline when the temperature
exceeds 800
0
e. This b e h a v i o ~ which has been observed with several other coals,
seems to arise from secondary reactions which are more pronounced at higher
temperatures and longer reaction times. It is also observed that the 80s yield
exceeds the 20s yield up to about 700
0
e where they become practically indisting-
uishable. This indicates that at 700
0
e (perhaps even lower) a pyrolysis time of
20s is sufficient for the attainment of the ultimate yield of tar.
The temperature-time behavior of the yield of hydrocarbon gases, also shown
in Fig. 4.9, is considerably different from that of tar. The ultimate yield
increases with temperature in the whole range studied. Moreover, the 80s curve
considerably exceeds the 20s curve up to about 900
0
e indicating the generally
higher activation energy of the respective rate determining steps.
Another detailed study of individual product yields was conducted by Suuberg
et al., also using the captive sample technique (refs. 63,64,71,72). Because
the experiments employed nonisothermal temperature pulses, the measurements do
not permit the ready visualization of the dependence on time and temperature,
although they are amenable to kinetic analysis as will be discussed in the next
chapter. Other results concerning product distribution in flash pyrolysis can
be found in refs. 65,67,73-75.
52
5.0
4.0
0 CO, He /3
u.
<[
+
Waler
0
0 Tm'M/20
0
0'-
o HCi3
{:, CO
.c.
3.0
'V Hydrogen
'"

Water
+
0

Tar'1120
m
2.0
.;:'
co
u

."
0
a:
1.0
Hydrogen
'Y

0 300 500 700 900 1100 1300
Pyrolysis temperature (OC)
Fig. 4.8. Product yields vs. temperature for a hva bituminous coal "Lower
Kittaning" at 20s pyrolysis time (source: ref. 70).
100
21
<[
0
,205
i!'
80
80S
18
U.
.c.
<[
'"
0
16
m
i!'

60

.c.
c
0
12
0
'"
0
"'
'"

0
0
u
0
u 40 9
."
m .c.
.;:'

0
i
f-
20
a
u
0;
0
;:
300 500 700 900 1100
Pyrolysis temperature
(OC)
Fig. 4.9. Yields of tar and gaseous hydrocarbons vs. temperature at
two pyrolysis times for a hva bituminous coal "Lower Kittaning" (source:
ref. 70).
53
4.2.3. Char and tar composition; distribution of sulfur and nitrogen.
In the previous subsection we presented data on the yields of tar, char and
several gaseous compounds. In this subsection we discuss the elemental composi-
tion and certain other properties of char and tar with an eye towards their
mechanistic significance. We also summarize data about the partition of sulfur
and nitrogen among tar, char and gases, which is important to the utilization
of these pyrolysis products as fuels.
Figure 4.10 and 4.11 display the elemental composition of char and tar as
functions of pyrolysis temperature for a high volatile bituminous coal studied
by Solomon (ref. 70). The effect of temperature on hydrogen, oxygen and carbon
is quite predictable. Increasing temperature is accompanied by a sharp decrease
in the fractions of oxygen and hydrogen due to the evolution of water, carbon
oxides and light hydrocarbons, all of which possess OIC or HIC ratios higher than
the parent coal. The sulfur content in the char is almost always lower than in
the parent coal. However, the temperature dependence of the sulfur varies with
the coal examined, probably due to the different amounts of inorganic and organic
forms. The nitrogen content in the char is somewhat higher than in the parent
coal.
The elemental composition of tar (Fig. 4.11) follows rather different trends.
Compared to the parent coal, tar is moderately enriched in hydrogen and sulfur,
considerably depleted in oxygen and approximately unchanged in carbon and nitro-
gen. With increasing temperature, hydrogen decreases slightly, while carbon remains
essentially constant. The temperature dependence of oxygen is erratic, perhaps
due to the error involved in determination by difference. With some exceptions,
the sulfur content is higher than in the parent coal but the temperature depend-
ence is erratic and varies from coal to coal. The content of nitrogen is gener-
ally similar to that in the parent coal and shows no noticeable temperature trend.
The distribution of sulfur in the pyrolysis products can be examined in more
detail in Fig. 4.12 (ref. 70) which gives organic and total sulfur normalized
with respect to the composition in the parent coal. All of the sulfur in the
tar is, of course, organic. The normalized organic sulfur is below one in both
tar and char, evidently due to the decompositon of reactive sulfur forms like
mercaptans and sulfides with the formation of gaseous products, mainly hydrogen
sulfide. In some coals with high pyritic content, the organic sulfur content
increases with temperature in both the char and tar due to incorporation in the
organic structure of decomposing pyritic sulfur. In other coals the organic
sulfur shows little temperature trend although it always remains below one. The
total sulfur in the char is somewhat less than unity and slightly declines with
temperature. References 70 and 76 by Solomon provide detailed results and dis-
cussion about the evolution of sulfur forms during pyrolysis.
54
.....
1.6
c

1.2
o
E

0.8
o
E

z
0.4
a Carbon
o Hydrogen
6 Oxygen
+ Nitrogen
V Sulf';lr
+
+ +_+_+ NitrOQen
n 0 "':--co,oo.
----===-0-
.::::::::::::::::::::;;::....oro--.;-- -------- Sulfur
Hydro;.n
OX)'Qen

Pyrolysis temperature (OC)
Fig. 4.10. Char composition vs. temperature for a hvc bituminous
coal "Ohio No.2" at 20s pyrolysis time (source: ref. 70).

1.6

0.8
o
E

z
0.4
o Corbon
o Hydrogen
A Oxygen
+ Nitrogen
V Sulfur
0 0
0 0 0 0
0
2
+__Hydrogen

+
*
:=a-e --=
-
i:I
'il

Oxy;en
l>
Nitrogen
l>

Pyrolysis temperature (OC)
Fig. 4.11. Tar composition vs. temperature for a hvc bituminous coal
"Ohio No.2" at 20s pyrolysis time (source: ref. 70).

5G
1.6
1.2
."
0.8
E
<;
z
0.4
o
e0:=:4=
o 0
o S(Ql chor
o SIT) chor
t::J. SIT) la,
+ SITlgos
V S(FeS.)
S(FeS.l
SlTlchor

8
Fig. 4.12.
coa1 (Ohi 0
o
Pyrolysis temperature (OC)
Distribution of sulfur in pyrolysis products of
No.2) vs. temperature at 30s; 0: organic, FeS :
x
iron, T: total (source: ref. 70).

a hvc bituminous
as sulfide of
===:::=:...>l.--,OQ....-<o,..':::"_-cor-->OL-...J0;L. To,
c 1.6

c
1.2

c
."
" 0.8
o
E
<;
z
o Char
o Tor
Gas
o
o
Q.-",,<:>---O--;:;8-- Chor
0.4
Go.

Pyrolysis temperature (OC)
Fig. 4.13. Distribution of nitrogen in pyrolysis products of a hvc bituminous
coal (Ohio No.2) vs. temperature at 30s (source: ref. 70).
56
Figure 4 . 1 ~ is an example of the nitrogen distribution in char, tar and gases
(ref. 70). The normalized nitrogen in the tar is close to unity indicating that
evolving tar molecules contain nitrogen functional groups in the same abundance
as in the coal. Char, on the other hand, contains a larger amount of nitrogen
while the gases a correspondingly lower amount.
The evolution of nitrogen at higher pyrolysis temperatures typical of pulver-
ized combustion has been studied by Blair et al. (ref. 67) and Pohl and Sarofim
(ref. 77). Figure 4.14 shows weight loss and nitrogen loss as a function of temp-
erature,clearly indicating the significant decrease in the nitrogen content of
char. Figure 4.15 shows the ultimate weight loss and nitrogen in the char as a
function of temperature for coals heated in crucibles. Although these experiments
involve slow heating they clearly show the nitrogen content to decline to almost
zero as the temperature increases from 1000 to 2000C.
The interpretation of the results shown in Figs. 14 and 15 is quite straight-
forward. Below 1000C nitrogen evolves only with the tar while the gases are
almost nitrogen free. As a result the char is gradually enhanced in nitrogen.
Above 1000C various forms of ring nitrogen in the char decompose with the evolu-
tion of products such as HCN resulting in a decrease of the nitrogen content.
In addition to the elemental analysis of char and tar, Solomon (ref. 78) and
Solomon and Colket (ref. 79) obtained Fourier transform infrared spectra and
13C_l H cross polarization nmr spectra. Figures 4.16 and 4.17 are reproduced from
this work. An inspection of Figure 4.16 shows the infrared spectra of coal and
tar to be remarkably similar indicating similar types and concentrations of bonds.
Figure 4.17 shows the nmr spectra of coal, tar and low temperature lOOOC) char,
in each case the broad peak representing aromatic and the narrow peak aliphatic
carbon. Coal and tar have identical values for the aromaticity fa while char a
somewhat higher value. The spectrum for the liquid product, tar, has considerably
more detail than the spectra for the solid materials.
Further characterization of the chemical structure of tar was carried out by
Gavalas and Oka (ref. 7). Three coals characterized in Table 4.1 were subjected
to pyrolysis under vacuum at 500C for 30s. The tar collected was separated by
gel permeation chromatography into three molecular weight fractions: L(>lOOO),
M(700-1000) and S(300-700). Each fraction was characterized by lH nmr and elemental
analysis. The results are shown in Table 4.2 along with corresponding results
for tetrahydrofuran extracts of the same coals.
A perusal of Table 4.1 shows coal to be closer in elemental composition to its
tar than its extract especially when the CO
2
evolved at 500C is deducted when
computing the elemental composition. Tar has essentially the same carbon content
as the coal but somewhat higher hydrogen and slightly lower oxygen contents. The
extracts on the other hand have substantially lower oxygen content.
1.0
Bituminous (Arkwright)
_.-0
--0
<e;"-..a 0 0
s->
0
I I
SUbbituminous (Wyodok)
.... 00
0-
0
.... a
0--
9- ........0
....
.... --
0
0.5
o
1.0
c
o
u
0.5
o
600 1000 1400 1800
Pyrolysis temperature ( C)
Fig. 4.14. Weight loss (----) and nitrogen loss
(-----) vs. temperature (source: ref. 67).
120
60
A,. Bituminous cool
0
100
50
.s::
u
c:
80
40
"0
'"
Ql
'"
c:
60
30
E
'0

:c
40
C'
c: 20
Ql
Ql

C'

20
10
Z
0
Fig. 4.15. Weight loss and nitrogen retention in char for
coal heated in crucibles (source: ref. 77).
57
3200 2400 1600 800
Wavenumber (em-I)
Fi g. 4.17. Infrared spectra of a hva
bituminous coal, its tar and low temp-
erature char (source: ref. 79).
ref. 79).
58
Cool
40
Cool
30
20
10
f
o
: 0.73
~
0
!!...
Tor
f1I
50
Tor
-
c:
40
:::I
>-
30
-
II)
f1I
<.J
20
c:
c:
II)
0
-
.0 10
c:
...
0
0
f1I
.0
ct
40 Low-T chor
Low-T chor
<5 (ppm)
Fig. 4.16. 13
C
_1
H
cross-polarization
spectra of a hva bituminous coal, its
tar and low temperature char (source:
T h ~ lH nmr data of Table 4.1 reveal that more than two-thirds of the hydrogen
is in aliphatic form. However, the split between H a ( a l p h ~ and HS+(beta or further
from the ring) varies with the rank of the coal. The H
S
+ predominates in the
subbituminous coal indicating long chains or hydroaromatic structures while
H
ar
+
o
' H
a
and H
S
+ are present in about the same amounts in the two bituminous
coals.
The molecular weight data of Table 4.2 show the middle fraction (M) to comprise
about half of tar and extract. At the same time, the heavy fraction (H) is more
abundant in the extract while the light fraction (L) is more abundant in the tar.
59
TABLE 4.1
E1 ementa1 composition of coa1s(daf), extracts and pyrolysis products (source:
ref. 7).
C H N S O(diff)
Subbituminous C, c 72 .1 5.00 1.28 0.85 20.7
Hyoming Monarch cc 75.2 5.35 1. 36 0.91 17.2
Seam, e 77 .8 8.75 1.28* 0.85* 11.4
PSOC 241
P
74.8 8.0 1.28* 0.85* 15.1
c 75.7 5.8 1.60 2.89 14.1
HVC Bituminous, cc 76.2 5.85 1. 62 2.93 13.4
Kentucky No.9 Seam e 82.1 7.4 1.60* 2.89* 6.0
p 76.1 7.1 1.60* 2.89* 12.3
c 85.9 5.75 1.48 0.74 6.2
HVB Bituminous, cc 86.0 5.75 1.49 0.74 6.0
H. Virginia e 87.5 7.3 1.48* 0.74* 3.1
P
85.9 7.25 1.48* 0.74* 4.7
c coal, dry-ash-free; cc coal after subtracting CO
2
evolved by pyrolysis
at 500C for 30s; e coal extracts; p coal pyrolysis products.
* Assumed for the purpose of estimating the oxygen content.
60
TABLE 4.2
Elemental analysis, molecular weights and nmr data for extracts and pyrolysates
(source: ref. 7).
Sub- HVC HVB
bitumi nous Bituminous Bituminous
Extract (wt %)a 4.5 8.0 1.6
Pyrolysis (wt %)b 6.0 19.1 10.5
GPC* (%)0
L 32 26 36 20 32 27
M 54 48 48 49 44 42
S 14 26 16 31 24 31
(%)d
H 20 23 32 34 32 37
1H nmr
ex
H
IH
68 58 40 33 37 31
H
ar+O
12 19 28 33 31 32
(%)e
13
C
nmr
C 42 58
ar
Cal
58 42
(%)
C 76.1 75.6 81. 3 75.1 85.7 85.8
L
H 8.9 8.1 7.8 7.1 6.9 6.7
C 77.5 73.7 81.5 76.5 87.6 85.9
M
H 8.7 8.3 7.1 7.1 7.6 8.0
C 82.6 76.1 85.5 - 89.5 85.9
S
H 8.5 7.3 7.4 7.2 6.7 -
a ~ O . 5 ; b.:t:.
l
. 5; c.:t:.l; d.:t:.2; e.:t:.2
* L,M,S: fractions with molecular weights >1000,700-1000, and 300-700
61
4.2.4 Effects of pretreatment and atmosphere of pyrolysis
Studies of pretreatment involve exposure to some agent at specified tempera-
ture and time, followed by pyrolysis in an inert gas, as usual. By limiting the
temperature and residence time, the extent of thermal decomposition reactions
during pretreatment can be minimized. Howard (ref. 80) has reviewed several
early publications on pretreatment by nitric oxide, steam and oxygen. Pretreat-
ment by nitric oxide at about 300
0
C (ref, 81) led to a modest loss of hydrogen,
a modest gain in oxygen and nitrogen and a substantial loss in the proximate
volatile matter. The swelling and agglomerating properties of the coal were
significantly reduced. Nitrous oxide had similar properties while nitrogen
dioxide reacted much faster, even at lower temperatures, and oxidized the coal
more extensively.
Pretreatment by steam has not shown any noteworthy effects. Heating a lignite
at about 300
0
C in the presence of steam (ref. 80) produced slight decrease of
weight loss and slight change of product distribution from the subsequent
pyrolysis.
Pretreatment in oxygen has long been studied as a means of reducing the
swelling and agglomerating properties of coal and is, in fact, a necessary step
in some of the gasification processes currently under development. Howard
(ref. 80) has summarized early studies showing that pretreatment by oxygen
increases the yield of carbon oxides and water formed in the subsequent pyrolysis
while it decreases the yield of tar and reduces or eliminates swelling and
agglomeration. The last effect was attributed principally to a change in the
surface of the coal which prevented particle agglomeration even when the particle
interior passed through a plastic phase. Forney et ale (ref. 82) found that
treatment with a gas containing Oe2% oxygen at 400-425
0
C for five minutes drasti-
cally reduced swelling and eliminated agglomeration of a caking coal. Under
these conditions, substantial devolatilization could not be avoided, McCarthy
(ref. 83) found that pretreatment in an atmosphere of 2-10% oxygen at 400
0
C for
a few seconds similarly greatly reduced agglomeration of a caking bituminous
coa1.
A comprehensive study of the effect of preoxidation on subsequent pyrolysis
and the properties of the resulting char was conducted by Mahajan et al. (ref.
84). A strongly caking coal was heated in air at 120-250
0
C for a few minutes
to four hours. The weight gain during this treatment increased with temperature
and time, not exceeding 6.5% under the most drastic conditions. Figure 4.18 shows
the weight loss of coal at different levels of prexidation as a function of
pyrolysis temperature. The level of preoxidation was in all cases fully charac-
terized by the percentage weight gain without reference to temperature and dura-
tion of the oxidative pretreatment. Below 450
0
C the level of preoxination has
62
40.-----------------,
30
-
:.e

CII
CII
E
20
- .t:.
co
CII

10

200 400 600 800 1000
Pyrolysi 5 temperature (OC)
Fig. 4.18. Pyrolysis weight loss vs. preoxidation
level for a caking bituminous coal psoe 337; weight
gain during preoxidation (%) none, 0 0.45,. 0.75,
o 1.4, &2.4,6. 7.0 (source: ref. 84).
little effect on the weight loss of pyrolysis but above 450
0
e the weight loss
decreases with the preoxidation level. The results shown in Fig. 4.18 are by
no means representative of all caking coals. Working with another caking coal,
the authors found a more complex dependence of the pyrolysis weight loss on the
level of preoxidation. At pyrolysis temperai:ures less than 500
0
e, the weight
loss increased the level of preoxidation but above 500
0
e the variation
became erratic. The complex dependence of weight loss was attributed to the
following competing factors. Addition of oxygen produces functional groups,
such as carboxyl, which during pyrolysis evolve carbon oxides and water making
a positive contribution to the weight loss. At the same time, the production of
carboxyl and other oxygenated groups reduces the amount of hydrogen that would
otherwise be available for tar generation. To this explanation, one must add that
increased water evolution during pyrolysis increases the degree of crosslinking
leading again to lower tar yield. In addition to the changes in the weight loss,
the chars of preoxidized coals had more open structure, sharply higher e0
2
sur-
face area, and moderately higher reactivity in oxygen.
The pyrolysis experiments discussed until now were all obtained either under
vacuum or under an inert gas such as nitrogen or helium. Pyrolysis in a hydrogen
atmosphere - hydropyrolysis - is important in its own right and will be examined
separately in chapter 7. Pyrolysis in an oxygen atmosphere is an important
63
part of combustion. It is generally believed that oxygen does not affect the
primary devolatilization reactions, although it rapidly oxidizes the volatiles
in the surrounding gas provided tiie temperature is sufficiently high as in a
combustion furnace. Only subsequent to the rapid volatile release, is oxygen
able to reach and oxidize til? cliQr particles. The effect of oxygen at lower
particle temperatures or when the surrounding gas is cold has not been studied
as such.
Pyrolysis in H
2
0 and CO
2
is of some interest but has received very little
attention. Jones et al. (ref. 85) compared the fluidized pyrolysis of a wet
(20% moisture) or predried subbituminous coal in nitrogen or steam with the
results shown in Table 4.3. The term "liquor" in the table indicates the
aquous phase containing the chemical water of pyrolysis along with small amounts
of phenolic compounds.
TABLE 4.3
Effect of moisture and fluidizing gas on product yields from the pyrolysis of
a subbituminous coal at 1400
0
F in a fluidized bed (ref. 85).
As received Dried at 300
0
F
Feed moisture % 20 20 0 0
Fluidized gas N
2
H
2
O N
2
H
2
O
Yields (% dry basis)
Char 54 0 6 53 0 56.0 53.1
Tar 1l.8 11.0 0.0 9.7
Liquor 7.0 8.7 9.6 11.4
Gas 26.6 27.3 25.4 25.8
The results of Table 4.3 indicate that the tar yield is slightly higher from
the moist (as received) coal irrespective of the fluidizing gas. The total
weight loss is slightly higher for the moist coal or when steam was the fluid-
lZlng gas. Although the differences are small and subject to some scatter, one
might venture the following conclusion. Drying coal induces changes in the
pore structure or the chemical structure that reduce tar and gas production in
subsequent pyrolysis. The observed increase in the yield of liqu"id (chemical
water) when steam is the fluidizing gas is rather puzzling and might be due to
errors in the material balances. The nature of the fluidizing gas seems to have
little effect on the yields of tar and gases suggesting little effect on secondary
tar-cracking reactions. Further discussion on the possible effects of H
2
0 or
CO
2
on secondary reactions will be given in a subsequent subsection dealing with
the pyrolysis process of the Occidental Research Corporation.
64
4.2.5 Effect of inorganic constituents or additives on pyrolysis product yields
Broadly speaking, inorganic matter can operate in two forms to influence the
pyrolytic or other reactions of coal. One is as discrete inclusions or continu-
ously distributed material within the coal particles. This form consists of
the inherent mineral matter of coal or of material added by impregnation or ion
exchange. The second form consists of mechanically mixed inorganic material
remaining external to the coal particles. The effects of these two forms of
inorganic matter will now be examined separately.
The main groups of minerals in coal include clays (e.g. kaolinite and illite),
silica (quartz), sulfides (mainly pyrite), carbonates (e.g. C a C O ~ ) , smaller
J
amounts of sulfates and oxides and minor amounts of other minerals. Detailed
composition normally refers to the ash, i.e. the inorganic matter remaining
after complete oxidation. The transformation of mineral matter to ash involves
loss of water from the clays, CO
2
from the carbonates, oxidation of pyrite to
iron oxide and fixation of oxides of sulfur in the form of sulfates by calcium
and magnesium oxides. The total amount of ash varies widely with coal but
generally remains below 25% by weight. The ash composition also varies widely.
For example, the composition of the major ash components in American bituminous
coals was found in the range SiO
Z
:ZO-60%, A1
Z
0
3
:10-35%, Fe
Z
0
3
:5-35%, CaO:1-Z0%,
MgO:O.3-4%, TiO
Z
:O.5-Z,5%, NaZO + K
Z
O:1-4% and S03:0.1-1Z% (ref. 86), Subbi-
tuminous coals and lignites contain larger amounts of CaO and MgO.
An important consideration relative to the catalytic activity of inherent
mineral matter in pyrolysis or gasification is size distribution. Mineral
matter is generally present in two forms, either as discrete particles of about
one micron size or larger, or distributed on a much finer scale, in association
with the organic matter. For example, calcium and magnesium in low rank coals
are largely present as cations associated with carboxylic groups. Mahajan
(reL 87) refers to measurements of N
Z
surface areas as hi gh as 10
2
m/ g for
coal minerals corresponding to a mean diameter of about 00 15 ~ m .
Detailed investigations of the effect of alkaline earth cations on the pyrol-
ysis products of low rank coals have been carried out by Schafer (refs. 88-90).
These studies compared acid-demineralized coals with coals that subsequent to
demineralization were converted to cation form (Na, K, Mg, Ca, Ba). The measure-
ments included the yields of CO
Z
' CO and HZO at different pyrolysis temperatures.
In addition, the carboxylic and phenolic content of the coal before and after
pyrolysis was determined by titration. The presence of catlons was found to
alter the relative yields of the three product gases but not to affect the
overall weight loss. For example, the ratio H
2
0/CO
Z
was always smaller for the
cation forms compared to the acid form. Significant differences in the gas
yields were observed between various cation forms, some possibly due to formation
65
of cyanide compounds by reactions of the cation with the carrier nitrogen. Com-
parisons between the CO
Z
' CO evolved and acidic content of the coal led to the
conclusion that CO
2
derives from carboxyl groups and CO derives from phenolic
Water was presumed to derive from some unidentified oxygen group
associated with carboxylic groups. These conclusions are at some variance with
the work of Brooks et al. (ref. 46) discussed in section 3,7. In a subsequent
investigation of the flash pyrolysis of various low rank coals Tyler and Schafer
(ref. 91) found that the presence of cations had profound effect on the yields
of tar, C
1
-C
3
hydrocarbons and total volatile matter. Removal of cations present
in the coals by acid wash increased the yield of tar by ao much as a factor of
two but had small effect on the yield of hydrocarbon gases. Conversely, addi
tion of calcium ions to the acid-form led to decreased tar and total volatile
matter. The fact that the change of the tar yield is not accompanied by a cor-
respondfng change of the yield of gases suggests that the effect of cations is
not manifested via secondary tar-cracking reactions. Instead it was suggested
that the cations might supress tar evolution either by restricting the micropore
structure, or by catalyzing the recombination of metaplast molecules before
volatilization could take place.
Mahajan and \Jalker (ref. 9Z) studied the effect of demineralization by acid
treatment on the N
Z
and CO
Z
surface areas of a number of coals and their carbon-
ization chars. Very divergent trends were observed among the various coals with
the surface areas in some cases increasing, in others decreasing and with N
Z
and
CO
2
areas not necessarily changing in the same direction. Changes in the porous
structure can certainly affect tar evolution but the evidence to date is clearly
insufficient for firm conclusions. The alternative explanation based on the
chemical role of mineral components is certainly plausible, especially consider-
ing that the acid treatment of coal to remove cations could also remove or
modify the acidity of clay minerals. Such clays might play some role in tar-
forming reactions via carbonium ion mechanisms. Further work to delineate the
pyrolytic behavior of cation exchanged coals is desirable especially in view
of the potential use of cations as catalysts or sulfur scavengers for gasifica-
tion and combustion.
The effect of inorganic additives in the form of powders mechanically mixed
with coal has also been examined. Yeboah et al. (ref. 93) studied the product
yields from a bituminous coal and a lignite pyrolyzed in a fluidized bed in the
presence of calcined dolomite particles, The of the dolomite resulted
in decreased tar yield and increased gas yield in all cases. These changes are
clearly due to secondary tar cracking reactions on the surface of the dolomite
particles.
A different mode of introduction of inorganic additives was investigated by
66
Franklin et al. (ref. 94). A Pittsburgh No.8 bituminous coal was demineralized
by extraction with a HF-HCl solution and co-slurried in water for for 24 hours
with extremely fine particles (0.1 of calcium carbonate or calcium oxide and
calcium carbonate. This pretreatment led to the incorporation of about 20%
calcium carbonate. While the form of the added material was not determined, it
probably consisted of particles coating the surface or penetrating the larger
pores of the coal with smaller amounts incorporated on a finer scale by associa-
tion with acidic functional groups of coal. The demineralized coal and the
calcium-treated coal were sJbjected to rapid pyrolysis by the captive sample
procedure with the results. Addition of calcium resulted in substan-
tially lower tar yield (20% versus 30%) and lower yield of light hydrocarbon
gases, especially at temperatures above 1100
0
K. At the same time, calcium
treated coals gave considerably higher yields of carbon monoxide, carbon dioxide
and water. The overall weight loss in the calcium treated coal was decreased
at temperatures above 900
0
K. The decreased yield of tar was attributed to
secondary reactions of cracking and repolymerization catalyzed by the calcium
additive. Such reactions would normally increase the yield of light hydrocarbons.
That the yield of these hydrocarbons actually decreased, could be explained by
the calcium additive catalyzing the further cracking of methane, ethane, etc.
to carbon and hydrogen. Much of the increase in the yield of carbon dioxide
was shown to result from the decomposition of calcium carbonate to calcium
oxide, a reaction catalyzed by the carbon surface. On the other hand, the in-
creased yield of carbon monoxide was attributed to the decomposition of phenolic
groups in the coal structure.
4.2.6 Miscellaneous techniques and results
In this subsection we review some additional pyrolysis studies which did not
fit properly in the previous sections, A number of early studies employed thermo-
gravimetric analysis with very low heating rates, a few degrees per minute. The
other group of studies employed heating by light, laser light, or plasma arc
achieving very high heating rates.
A thermal balance is an apparatus providing continuous measurement of the
weight of a static sample under the flow of a carrier gas and linearly rising
temperature. Waters (ref, 95) described some early thermal balances and dis-
cussed weight loss curves from coal pyrolysis, van Krevelen et al. (ref. 96)
used a thermal balance of the torsion type to correlate weight loss with plastic
properties. Figure 4.19 shows some typical weight loss curves obtained with a
low volatile bituminous coal. The chief features of these curves are (i) The S-
shape of the cumulative weight loss. The initial acceleration of the curve,
observed also in section 4.2.1, could be due to consecutive reactions or the
solubility of some pyrolysis products in the coal melt. (in The temperature
67
of maximum devolatilization rate increases with increasing heating rate. More
recently, Ciuryla et al. (ref. 97) employed a modern thermobalance to study the
weight loss of coals of different ranks under heating rates 40-160
0
/min. As
in earlier studies, maximum rate of devolatilization increased with heating
rate. The cumulative yield ata final temperature of laaa
o
C was found inde-
pendent of the heating rate, in agreement with the studies reviewed in the
previous subsections. The weight loss data were fitted successfully by the
Pitt-Anthony model of distributed activation energies.
The principal advantage of the thermogravimetric technique is the continuous
and accurate weight measuremento Its chief disadvantages is the inability to
operate at high heating rates and constant temperatures and the difficulty to
measure accurately the sample temperature. Isothermal operation must always be
II 0.11

9


III 7

......
III
:ae E
0.6
5
0.8
--
-
1.2 c:
.s:::.
2.5
III
0>
4.9
III
"Qi
3 6.8
0
--
:t
C/min
:!::
a
0.01
400 500 600 400 500 600
Temperature
(Oe l
Temperature
(Oel
Fig. 4.19. Thermogravimetric analysis of a 10 volatile
bituminous coal. Cumulative (a) and differential (b)
weight loss temperature at different heating rates
(source: ref. 96).
preceded by a period of slow linear heating. With current commercial models,
the sample size can be varied from a few milligrams to about a gram. With large
sample size, the sample is uncertain and secondary reactions may
become important. Using a small sample size and a sweep gas minimizes secondary
reactions but renders product analysis difficult. In fact, none of the afore-
mentioned studies measured individual product yieldso Combination of continuous
and accurate weight measurement with continuous analysis of gaseous products by
infrared and flame ionization detectors seems a most promising technique despite
the restriction to low heating rateso
The kinetic feature of constant and slow heating rates has been employed
without the gravimetric capability, simply by placing the coal sample in a
furnace under the flow of a sweep gas. Using a suitably large sample, of the
68
order of one gram, it is relatively straightforward to measure the instantaneous
rate of generation of individual gaseous species at the cost, of course, of
allowing secondary reactions on the particle surface. Experiments of this type
have been carried out by Fitzerald and van Krevelen (ref. 98) and Juntgen and
van Heek (ref, 66).
Investigation of pyrolysis in the context of pulverized combustion requires
very high heating rates in the range of 20,000 - 100,000 C/s. One experimental
technique for achieving such rates in a small scale apparatus is the irradiation of
coal with ordinary or laser light. Sharkey et al. (ref. 99) irradiated coal
in the form of 3/8" cubes or a fine powder with light from a xenon lamp or a
ruby laser operating at specified power levels. The coal temperature attained
under these conditions could not be determined but was estimated to be in excess
of 1000
0
C. The product yields obtained under irradiation compared to those
obtained under ordinary carbonization showed much higher contents of acetylene
(absent in carbonization) and carbon oxides and sharply lower content of methane.
These results were attributed to the higher temperatures prevalent under irradia-
tion" However, the yields of hydrocarbons other than methane and acetylene
exhibited irregular and on the whole obscure variations. Evidently, secondary
reactions in the gas phase, as well as in the solid phase, are responsible for
the overall product distribution.
Granger and Ladner (ref. 100) analyzed the gaseous products of several coals
under irradiation from a xenon lamp with and without a filter to remove the UV
component of the light. Their small scale apparatus allowed variation of the
incident energy and their analysis included gaseous products, i.e. carbon oxides
and light hydrocarbons. Water was not determined and tar was estimated rather
crudely. Figure 4.20 shows the major pyrolysis products versus total incident
energy. The principal feature of the results is the rapid increase in hydrogen,
acetylene and carbon monoxide with increasing light intensity. The rapid
increase in carbon monoxide is easily accounted by the fact that this gas is the
only oxygen-containing product after the evolution of carbon dioxide has been
completed in the initial phases of pyrolysis. The latter gas was actually
absent, evidently being converted to carbon dioxide and carbon at the high tem-
peratures of the experiment. The yield of tar was found to pass through a
maximum as a result of secondary cracking reactions. The results of Fig. 4.20
were obtained using UV-filtered radiation, Unfiltered light resulted in decreased
yields of olefins and paraffins and increased yields of carbon monoxide, hydrogen,
acetylene and soot. These changes indicate the presence of photochemical crack-
ing reactions in the gas phase.
A technique for heating coals at very high temperatures and heating rates is
the plasma arc. Bond et al. (ref. 101) used an argon plasma jet to pyrolyse
69
1.5
~
::.e
0
-
.c.
1.0
C'
CI>
~
"00.5
CI>
>-
o~ l I I : I I I I ~ ~ ~ - - - - : l l : - - - : l .
o 20 40 60 80 0 20 40
Incident energy / flash (kJ/m
2
)
Fig. 4.20. Yields of gaseous products from the flash
pyrolysis by xenon light of a noncaking (a) and
a caking (b) coal (source: ref. 100).
coals of various ranl:s. The device provided rapid heating and quenching of the
pulverized coal particles, Temperatures at the center of the jet were estimated
to be between 10,000 to 15,0000C. The maximum temperature attained by the coal
particles could not be determined but was estimated to be well in excess of
10000C. The products were gas and soot with tar completely absent. The gas
contained mainly hydrogen, carbon monoxide and acetylene, which comprised over
95% of all hydrocarbon gases. Acetylene yields as high as 20% by weight of
the coal were obtained. Overall conversion increased with decreasing particle
size and with increasing proximate volatile matter of the starting coal.
A thermodynamic analysis of product distribution from ultrahigh temperature
coal pyrolysis in various atmospheres was made by Griffiths and Standing (ref.
102). Although, in each case the equilibrium mixture contains free radicals as
well as stable species, attention can be limited to stable species assuming
suitable radical recombination during quenching. At temperatures above 1800
o
K,
acetylene is the most stable among hydrocarbons, although unstable with respect
to carbon and hydrogen. Above 3,OOOoK, the yield of acetylene at equilibrium
with carbon and hydrogen becomes significant. The oxygen present in coal probably
ends up as carbon monoxide, however, oxygen was not considered. In the presence
of nitrogen, the equilibrium yields of cyanogen and hydrogen cyanide are also
significant. The actual product yields obtained in a practical configuration
such as a plasma arc probably do not reach thermodynamic equilibrium because
of insufficient residence times.
70
4.3 PYROLYSIS PROCESSES
Pyrolysis is 'the simplest means of upgrading coal to higher quality fuels.
Merely by heating, coal decomposes to gases, tar liquids and char. The gases
can be readily burned in an industrial furnace. The tar is the most valuable
product because it can be hydrotreated to clean liquid fuels. The char must be
utilized in an industrial or utility furnace or gasified to a low Btu or synthesis
gas. Because of its low content of volatiles char has poor ignition character-
istics and may' require special burners or some other means of maintaining flame
stability. It can always be burned in a fluidized furnace.
The simplicity of the basic flow sheet of pyrolysis as a coal conversion proc-
ess obscures a number of mechanical difficulties that have slowed down its commer-
cial development. Chief among these difficulties is the caking and agglomer-
ating properties of high volatile bituminous coals upon heating. Those very
coals are also the most interesting for their high yield of tar liquids. Rapid
heating of coal in a configuration that limits the extent of secondary reactions
is another difficulty that has not been entirely overcome. Finally,collecting
tar liquids and removing suspended fine solids is also a problem, common to other
coal conversion processes. The two processes discussed below have at least par-
tially overcome the mechanical problems associated with rapid heating and agglom-
eration.
4.3.1 The COED process (refs. 103-105)
The COED process (Char-Oil-Energy-Development) was developed by FMC corporation
under contract from the office of coal research (subsequently absorbed into ERDA
which in turn was absorbed into the department of energy). The development effort
reached the stage of a 36 ton-per-day pilot plant tested in the period 1971-1973.
Since then, research and development activities were redirected to a related gas-
ification process (COGAS) under private sponsorship.
The central part of the process is shown in the schematic diagram of Fig. 4.21.
Coarsely ground coal 1 mm diameter) is dried and fed to fluidized bed I operating
at 600F with hot combustion gases. The volatiles from I flow to the product re-
covery system while the char is carried to fluidized bed II operating at 850F and
subsequently to fluidized bed III operating at 1000F. Char from stage III is
carried to fluidized bed IV (1600F) where it is partially burned with oxygen.
Hot char from stage IV is recycled to stage III to provide part of the required
heating. The effluent gases from stage IV provide additional heating as well as
fluidizing for stage III. The hot product gases from stage III in turn provide
the heating and fluidizing medium for stage II. The gaseous and liquid products
from stages I and II are separated. Part of the gases is burned to provide the
fluidized medium for stage I, the remainder collected as an end product of the
71,.
recycle gas
5to gel 1-+-"";'-,0,,,-,",,-'-"Ii",o,,,,oc_.J Pra duet
Recovery gas
cool
'----------oil to hydrotreoting

steam
product char
oxygen recycle gas
from Stage I
product recovery
Fig. 4.21. The COED process flowsheet (source; ref. 103).
process. The product liquids are hydrotreated to clean liquid fuels for station-
ary or portable powerplants.
Carrying out the pyrolysis in four coupled fluidized beds provides the re-
quired heating and at the same time prevents caking and agglomeration. As coal
(or char) progresses through stages I-IV its caking temperature increases due to
the successive loss of volatiles. Thus maintaining the temperature of each stage
lower than the caking temperature of the fluidized char prevents its softening
and agglomeration. On the other hand, the prolonged contact of volatiles with
the fluidized char result in extensive secondary reactions. Another adverse
factor in terms of secondary reactions is the relatively large particle size.
Compared to the tars produced in the laboratory reactors described in section 4.1,
the COED liquids are produced at lower yields and have lower boiling points and
1ess polar character, whence the term "oil s"
A ton of high volatile bituminous coal treated by the COED process yields
about 1.4 barrels of oil or about 18% by weight, well below the 30-50% laboratory
yields by the captive sample technique. The other products from one ton of coal
are char, about the same as the ASTM proximate analysis, and 8000-9000 scf of
gases of heating value about 540jscf. The relatively low yield of liquid products
72
is probably the main reason why the COED process has been dropped from the small
list of liquefaction processes scheduled for commercialization.
4.3.2 The Occidental Research Corporation (ORC) process (ref. 106)
The ORC process has been under development since 1969. The original experi-
mental work at the laboratory scale was internally funded. Subsequent work util-
ized the laboratory unit and a three ton-per-day process development unit which
was tested in the period 1976-1978 under contract to the department of energy.
A schematic of the ORC process flowsheet is shown in Fig. 4.22. Coal is pul-
verized (median size 20-40 ~ m ) , dried and pneumatically transported to the pyroly-
sis reactor where it is mixed with hot recycle char. Solids and gases move co-
currently downward through the reactor and are collected in a cyclone. The solid
stream from the cyclone is split in two parts. One part is carried to an entrained
flow vessel for partial burning and recycling to the reactor. The remainder is
removed as the product char. The volatiles from the cyclone are rapidly quenched
and separated into a fuel gas and a liquid product which after hydrotreating pro-
vides the main process product .
.---------------.Combustion Gas
Air
Cool
Feed
Char
Burner
,-----Gas
Oil Collection System
Fuel
I---Char Product
Desulfurization Plant
Fig. 4.22. Schematic of the ORC pyrolysis process (source: ref. 106).
73
Some of the key operating parameters in the reactor are temperature (1000-
l400F), pressure (5 psig) and ratio of recycled char to coal, about 10: 1. The
results from the bench scale reactor (BSR) and the pi 1ot plant or "process de-
ve1opment uni t" show some important di fferences and wi 11 be di scussed separately.
The bench scale unit consisted of an externally heated entrained flow reactor.
Recycle char was not used in this system since the main objective was to deter-
mine the dependence of product yields on temperature and residence time. Figure
4.23 shows the yields of the major pyrolysis products as functions of temperature
for two residence times, 1.5s and 3s. The mcst interesting feature of the figure
is the maximum in the tar yield at about 1200"F. The maximum yield of about 20%
is twice the amount obtained in the standard Fischer assay. As the temperature
increases beyond 1200"F the tar yield declines, evidently due to secondary re-
actions occurring homogeneously or on the reactor walls.
90,-----,-----r--..----,----.----,
-- 1.55
---3.05
......
............
....
.....
.........
....
"
"
.............. Char
'..
80
10
u
OJ 20
-0
2
a..
LL
<J:
o 70
if!.
.<: 60
0'
Q)
3
- 50
-0
OJ 30

900 1100 1300 1500
Reactor temperature (O F)
Fig. 4.23. Product yields vs. temperature from
the pyrolysis of a subbituminous coal in the BSR
(source: ref. 106).
74
Pyro1yzing the subbituminous coal in the process development unit under simi-
lar operating conditions but with hot char recycle gave a surprisingly different
product distribution. The tar yield was limited to only about 7-10%, while char
and gases were produced in increased quantities. Since the only essential diff-
erence between the operation with the BSR and PDU was the presence of the re-
cycle char, the difference in the tar yields was attributed to adsorption and
cracki ng on the char surface. To confi rm thi s exp1 anati on additi ona1 experiments
were performed with increased amounts of recycle char. Changing the char to coal
ratio from 10:1 to 40:1 was found to decrease the tar yield from 10% to 4%.
Continuing efforts to the tar yield resulted in an unexpected finding.
When the inert N
2
carrier was replaced by CO
2
or H
2
0, the maximum tar yield from
the PDU operation increased to 18-22%, a level identical to that obtained with
the BSR. This important and surprising result was attributed to the adsorption
of CO
2
and H
2
0 on the active sites of the char's surface which would otherwise
catalyze the cracking of tar molecules. In other words, CO
2
and H
2
0 compete
with tar for the same active sites on the char's surface.
40
I.L
<t
a
30
::!
0
-.z=
C'
Q)
20

"0
Q)
>-
...
10
0
I-
0
0 5.0
Residence
Fig. 4.24. Tar yield vs. residence time from the pyrolysis of a
bituminous coal in the PDU (source: ref. 106).
75
The results discussed up to this point concern the pyrolysis of a subbituminous
coal. PDU experiments were also performed with a high volatile bituminous coal
(hvc, Kentucky No.9). Figure 4.24 shows the yield of tar as a function of temp-
erature at two residence times. The maximum yield of about 38% would be quite
attractive on a commercial scale. The decline of the yield with increasing res-
idence time is again due to secondary reactions. Surprisingly, the tar yield for
the bituminous coal was found to be essentially independent of the carrier gas
(N
2
,C0
2
,H
2
0). The strikingly different behavior of the two coals has not been
explained and certainly deserves systematic study.
At this point we return to an earlier observation that the caking and agglom-
erating properties of coal constitute the chief technical difficulty in the
commercialization of pyrolysis. The ORC process has approached the difficult
problems of coal agglomeration by a special design of the top part of the PDU
reactor where the mixing of coal and recycle char takes place (see Fig. 4.25).
Pulverized coal is injected by a jet of carrier gas in a surrounding stream of
hot recycle char descending downwards from a shallow fluidized region. It is
vital that the coal particles are heated rapidly to complete their caking phase
before reaching the reactor walls. Small particle size and large char to coal
ratio are necessary for this purpose.
Char
t
1::::::::;==-Fluidizing Gas
Fig. 4.25. Device for injecting and mixing of coal and char
in the PDU (source: 106).
76
The reactor ,design shown in Fig. 4.25 was only partially successful. During
operation with the bituminous coal the PDU encountered flow instabilities limit-
ing continuous operation to less than one day. A momentary decrease in the char
flow rate would cause the deposition of plastic coal particles on the injector
tip or the reactor walls. The resulting decrease in the cross section would in
turn further reduce the char flow rate and so o ~ leading eventually to a complete
flow stoppage. Although improved designs evolved during the testing, a completely
satisfactory operation was not achieved.
Assuming that the mechanical difficulties will eventually be resolved, the
ORC process appears a most promising route to coal liquids by virtue of its
simple flowsheet and the high yield of tar.
77
Chapter 5
HEAT AND MASS TRANSFER IN PYROLYSIS
The experimental data reviewed in the previous chapter generally referred to
small particles 100 ~ m ) and low pressures (atmospheric or lower). Under these
conditions, heat and mass transfer are rapid and have relatively small influence
on weight loss and product yields. In this chapter we will examine the more
general situation in which mass and, to a smaller degree, heat transfer have a
significant influence on product yields. In Chapters 2 and 3 we discussed
chemical structure and reactions with minimal reference to physical properties
such as viscosity and porosity. These properties have a decisive effect on
transport phenomena. Thus in the first section, we briefly discuss the relevant
physical properties of plastic and nonplastic coals. In the second section we
survey experimental data on the effect of pressure and particle size which, of
course, reflect the presence of transport limitations, and in the last section
we develop a simple theoretical treatment of these phenomena. Although heat
and mass transfer are coupled to the kinetics of pyrolysis, the scope of the
theoretical analysis will be limited to problems that can be treated without
reference to detailed kinetics.
5.1 PYROLYSIS AND THE PHYSICAL PROPERTIES OF COAL
5.1.1 The plastic state of coals
The two physical properties that govern the rate of transport processes,
especially mass transfer, are the viscosity during the plastic stage, and the
porous structure of coal. These two properties are not independent because the
plastic properties determine to a major degree the evolution of the porous
structure during pyrolysis. Under certain conditions, when heated above about
350
0
C coals melt to a highly viscous, non-newtonian liquid, or melt, whence the
term "plastic", or "softening" coals. Whether or not such melting or softening
takes place and the actual viscosity or fluidity (reciprocal of viscosity) of
the coal melt depend on rank, heating rate, particle size, pressure and surround-
ing gas.
Almost all rheological measurements of coals have been conducted at low
heating rates - a few degrees per minute - using a few grams of coal sample.
Under these conditions, certain coals, chiefly bituminous, become fluid. Fluidity
is most evident in coals with carbon content (dry, ash-free) in the range 81-92%
with a maximum at about 89% (ref. 107). Carbon content does not, of course,
provide complete characterization of rheological properties. The contents of
hydrogen and oxygen are also strongly correlated with fluidity. At fixed carbon
78
content, fluidity decreases with increasing oxygen content. The latter property
has been already discussed in section 4.2.4.
The conditions of heating play an important role in the development of plastic
properties. At fixed heating rate, increasing pressure and particle size or mass
of sample result in increased fluidity. Particle size and heating rate can, of
course, be varied independently only within a limited range. The plastic proper-
ties of dilute pulverized particles depend strongly on the heating rate. Hamilton
(refs. 108, 109) heated dispersed vitrinite particles (100 ~ m ) to 1000
0
C in nitro-
gen employing heating rates in the range 10-
1
to 10
4
C/s and observed manifesta-
tions of plastic behavior such as the rounding of the particles and the formation
of vesicles and cenospheres. He found a striking relationship between coal rank
and heating rate required for plastic behavior. For high volatile bituminous
coals, changes in char morphology such as rounding, vesiculation, etc. started
becoming significant at about 1 C/s and increased up to about 10
2
C/s. Further
increase of the heating rate beyond 10
2
C/s did not produce any further morpho-
logical changes. Vitrinites of lower or higher rank e.g. subbituminous and
semianthracites required heating rates of 10
2
C/s or higher before they displayed
any morphological changes. Once manifested, such changes increased up to 10
3
to
10
4
C/s, depending on the particular vitrinite. These results suggest that
coals of different rank can be made to exhibit similar plastic behavior by
suitably adjusting the heating rate.
The plastic behavior of coal and its dependence on rank and heating rate can
be qualitatively accounted by the reactions of pyrolysis. With increasing temper-
ature, the disruption of secondary bonds and the dissociation of covalent bond
induces melting and fluid behavior. The extent of covalent bond breaking required
for this purpose is probably limited, at least for high volatile bituminous coals.
Concurrently with bond breaking, other processes work in the opposite direction
to increase molecular weight and resolidify coal. These are the loss of tar,
which increases the average molecular weight of the remaining material, and the
free radical recombination and various condensation reactions (e.g. condensations
of phenolic groups) which also increase molecular weight. The balance of these
processes determines the occurrence, extent and duration of the fluid or plastic
state. Anthracites are too heavily graphitic in character to exhibit plastic
behavior. Semianthracites and low volatile bituminous coals contain highly con-
densed aromatic units of relatively large molecular weight. They exhibit some
plastic behavior at sufficiently high heating rates. High volatile bituminous
coals consist of units of lower molecular weight and exhibit maximum fluidity.
With further decreases in rank, the molecular weight of the starting material
would not necessarily decrease, but increased polarity and condensation of phenolic
groups restrict the range and extent of plastic behavior. In particular, plastic
79
properties are exhibited only at high heating rates. Under such rates, however,
fluidity commences at high temperatures and is of short duration due to the
acceleration of all reaction rates.
Almost all measurements of rheological properties of coal have been conducted
at heating rates of a few degrees per minute. At such rates only bituminous
coals exhibit fluid behavior, the fluidity commencing just below 400
0
C, the
maximum fluidity being attained at about 450
0
C, with resolidification taking
place above 500
0
C. The resolidification relates to the essential completion
of tar evolution and the increased molecular weight of the residual material.
Coals that exhibit pronounced fluid behavior, are commonly called softening or
plastic coals.
The transformation of a coal to a liquid and its subsequent pyrolytic decom-
position induce physical changes that have a profound effect on the transfer of
pyrolysis products. Following melting, preexisting pores partly collapse due to
surface tension forceso The volatile products of decomposition initially dis-
solved in the melt start nucleating once their concentration exceeds a critical
level and the nuclei formed coalesce into larger bubbles which eventually break
through the particle surface. Nucleation, growth and bursting of bubbles consti-
tute the chief route of intraparticle mass transfer.
The formation and growth of bubbles causes an expansion or "swelling" of the
coal particles. The degree of this swelling depends on particle size, external
pressure and heating rate or, generally, temperature-time history. Swelling
factors (volumetric) as large as 25 have been observed (ref. 110) under rapid
heatinq conditions. To characterize the swelling properties of coals a standard-
ized test has been developed providing the "free swelling index".
The viscosity of coal in its plastic state has a pervasive influence in many
rate processes of interest. It regulates the dynamics of nucleation, bubble
growth and bubble coalescence and has an inverse relationship with the diffusion
coefficient of pyrolysis productso It also affects the intrinsic kinetics by
controlling the rate of bimolecular reactions such as free radical recombination.
Viscosity or fluidity (the reciprocal of viscosity) is a transient property and
comparisons among different coals are meaningful only under specified conditions
of temperature time history, particle size, etco
Waters (ref. Ill) has made extensive rheological measurements suggesting a
close relationship between fluidity and instantaneous weight loss. This rela-
tion is, of course, due to the fact that fluidity and devolatilization must both
be preceded by covalent bond breaking. The rheological properties of coal can
be measured by several techniques which have been reviewed in the monograph of
Kirov and Stevens (ref. 112). The most common of these techniques employs a
rotational viscometer known as the Giesel plastometer, which measures the
80
"fluidity" of coal as a function of time at a heating rate of 3
0
C per minute
and other specified experimental conditions.
At heating rates characteristic of flash pyrolysis (several hundred or thous-
and degrees per second) it is not possible to measure the viscosity, although it
is still possible to observe swelling and bubble formation. At high heating
rates, the inception of fluidity, the point of minimum viscosity, and the
resolidification are displaced towards higher temperatures, in close relation
with the rate of weight loss.
An important consequence of coal's plastic properties is the agglomeration
of particles to grape-like structures Ot' to a completely coalesced mass or "cake"
whence the terms "agglomerating" or "caking" are often used in place of "softening".
By contrast, coals which exhibit a limited range of fluid behavior (e.g. sub-
bituminous and lignites) are normally considered as "nonplastic", "nonsoftening",
"noncaking", or "nonagglomerating". The agglomeration of coal particles is a
serious difficulty in the operation of fixed bed or fluidized bed gasifiers and
has also been identified as the most serious technical obstacle in the development
of commercial pyrolysis processes as a route to coal liquids (section 4,3).
5.1,2 Changes in the porous structure of coal during pyrolysis
The pore structure of coals has been comprehensively treated by Walker and
Mahajan (ref. 113) and more recently by Mahajan (ref, 114). These references
discuss experimental techniques available for the measurement of surface area,
pore volume and pore size distribution. In this subsection we will summarily
review the aspects of coal porosity that have an important bearing on transport
processes during pyrolysis and the changes of porous structure occurring during
pyrolysis.
Coals have a very complex pore structure, both in terms of size distribution,
which is very broad, and in terms of the geometry of individual pores or voids.
Following ref. 115, we classify pores according to pore diameter into micropores:
0.4 - 1.2 nm, transitional: 1.2 - 30 nm and macropores: 30 - 1000 nm.
Table 5,1 below reproduces measurements of Gan et al. (ref. 115) of pore
volumes in the three size ranges for several American coals, The total pore
volume V
T
was computed from helium and mercury densities, the macropore volume
VI was estimated from mercury porosimetry, the transitional pore volume V
2
was
estimated from the adsorption branch of the nitrogen isotherms and the micropore
volume was estimated by difference, V
3
=V
T
-V
1
-V
2
,
The last two columns in table 5.1 list the surface areas obtained by adsorption
of nitrogen and carbon dioxide, SN2 was calculated using the BET equation while
SCO was calculated using the Dubinin-Polanyi equation. The difference between
t h e ~ e two areas has always been of great interest. It is generally attributed
81
to the ability of the carbon dioxide molecule at 298
0
Kto penetrate pore openings
0
as small as 4 A, whereas the slightly smaller nitrogen molecule at 77
0
Kcan only
penetrate openings larger than about 5 E.
TABLE 5.1
Pore volumes and surface areas of several American coals (ref. 115).
Designation Rank
3
V
1
(%) V
2
(%) V
3
(%)
2 2
VT(cm /g) SN (m /g)
SCO (m g)
2 2
PSOC-80 Anthr. 0.076
r ('
13.1 11.9 7.0 408 :l.U
PSOC-127 lvb 0.052 73.0 0 27.0 <1 .. 0 253
PSOC-135 mvb 0.042 61.9 0 38.1 <1.0 214
PSOC-4 hvab 0.033 48.5 0 51.5 <1.0 213
PSOC-105A hvbb 0.144 29.9 45.1 25.0 43.0 114
Rand hvcb 0.083 47.0 32.5 20.5 17.0 147
PSOC-26 hvcb 0.158 41.8 38.6 19.6 35.0 133
PSOC-197 hvcb 0.105 66.7 12.4 20.9 8.0 163
PSOC-190 hvcb 0.232 30.2 52.6 17.2 83.0 96
PSOC-141 1ignite 0.114 19.3 3.5 77,2 2.3 250
PSOC-87 1i gni te 0.105 40.9 0 59,1 <1.0 268
PSOC-89 lignite 0.073 12.3 0 87,7 <1.0 238
Consider for example sample PSOC-26. The surface area of micropores includes
that of pores with openings below 5 ~ is S' =SCO - SN = 98 m
2
/g. From S' and
V
3
= 0.03 cm
3
/g we can estimate a lower bound for
2
the m ~ a n size of the micropores.
Assuming spherical shape, the mean diameter of micropores must be at least
6V
3
/S' = 1.8 nm. This estimate suggests that the microporous space largely con-
sists of voids having diameter of a few nm which, however, are accessible via
much smaller openings. This particular feature of the microporous system, the
"aperture-cavity" structure, has been poi nted out by many authors, e. g. refs.
116, 117. Similar observations have been made for cokes (ref. 113) from carbon-
ized coal.
The coals listed in table 5.1 contained substantial porosity in the micro
and macro ranges. In particular, micropores constituted more than 60% of total
volume in the high rank coals. By contrast, only the high volatile bituminous
coals had significant volume in the transitional pore range. Figure 5.1 shows
the cumulative pore volume distribution of one such coal (PSOC 190). The dis-
tribution covers a wide size range from a few angstroms to about one micron.
The changes in the pore size distribution accompanying pyrolysis depend a
great deal on the rank of the coal and, in particular, on its softening or plastic
82
properties. It is thus essential to distinguish between softening and nonsoftening
coals. Nsakala et al. (ref. 119) measured the He and Hg densities and the N
2
and
CO
2
surface areas of two lignites as a function of isothermal pyrolysis time at
800
0
C. The lignite particles were injected with a stream of preheated nitrogen
in a vertical furnace thus achieving heating rates about 10
4
C/s. They also
measured the He and Hg densities for slow heating (lOoC/min) in a fluidized bed
maintained at 800
0
C. At the high heating rates, the helium density increased
while the mercury density decreased with pyrolysis time so that the total open
pore volume given by
1
V=-"--
T il
Hg
increased with pyrolysis time. The slow heating produced negligible change
in the mercury densities but substantial increase of the helium densities. As
a result, the helium and mercury densities of the chars produced under slow
heating were larger than those possessed by the chars produced by rapid heating.
Rapid heating produced sharp increases in the N
2
and CO
2
surface areas as shown
for one of the two lignites in Fig. 5.2. In this case, the N
2
surface area
increased by a factor of almost one hundred while the CO
2
surface area almost
doubled.
2L.....__.L.-__.L.-__.L.-_--J
10
o 0.05 0.10 0.15 0.20
Cumulative pore volume (cm
2
/g)
c:
Fig. 5.1. Cumulative pore volume distribution of a hvc
bituminous coal "Illinois No.6" (source: ref. 115).
83
The authors discussed their experimental results in terms of two competing
processes. Thermal bond breaking produces tar and other volatiles, the removal
of which increases pore volume and widens constrictions, whence the large increase
in the CO
2
surface area. Bond breaking also facilitates the alignment and coales-
cence of coal's structural units tending to decrease pore volume and surface area.
At the same time, bond formation or cross-linking results in decreased pore
volume and surface area. The balance of these processes depends on coal rank,
heating rate, maximum temperature, and time at the maximum temperature. For
lignites rapidly heated to 800
0
C, volatile removal predominates over alignment
and cross-linking leading to increased open pore volume and surface area. In
this respect, the increase in the He density in conjunction with the sharp in-
crease in surface area probably signifies the widening of previously impenetrable
apertures. The decrease in the Hg density reflects the removal of material which
at high heating rates is not accompanied by compensating particle shrinkage, At
slow heating, volatile removal is evidently supplemented by cross-linking and
alignment leading to much higher He densities but leaving the Hg densities un-
affected.
Nandi et al, (ref, 120) measured changes in the pore volume and surface area
of three anthracites upon pyrolysis at different final temperatures with heating
rates of 5
0
C/min. The helium densities in all cases increased with final tem-
perature to about 20% above their initial value. The changes in the mercury
400,--------------..,
-.
-
o
'0
C;;300
.....
...
E
200....._-
o
..
...
o
oI.::tt:=tl:::::::.L_--L_..l.----.J
o 0.2 0.4 0.6 0.8 1.0
Time (sec)
~ 100
o
-
...
::J
en
Fig. 5.2. N
2
and CO
2
surface areas of a lignite as a function
of residence time in a vertical furnace. Large times achieved
by multiple passes (source: ref. 119).
84
densities were smaller and had no definite direction. The change in the total
open pore volume was also small and erratic. The N
2
and e0
2
surface areas of
two of the three anthracites increased with temperature, passed through a
maximum at about GOOoe and then decreased sharply above 800
0
e. The areas of the
third anthracite declined slowly until about 800
0
e and rapidly thereafter. The
increase in the surface areas at the lower temperatures can again be attributed
to the loss of volatiles (e.g. carbon oxides) widening the micropore openings.
At the higher temperatures, cross-linking between adjacent units decreased
micropore openings causing the sharp decline in surface area.
Toda (refs. 121, 122) studied changes in the pore structure of several Japanese
coals following a treatment consisting of heating at 3
0
e/min to a final temper-
ature and holding at that temperature for 15 min. The pore structure was probed
by mercury porosimetry as well as by measuring the densities in helium, methanol,
n-hexane and mercury. Figures 5.3 a, b show the specific volumes in mercury and
n-hexane as a function of the final pyrolysis temperature for a nonsoftening and
a softening coal respectively. The specific volume in n-hexane is the volume
impenetrable by n-hexane, while the volume in mercury is the total particle volume.
The difference between the two is the total volume of pores with size above a few
angstroms, i.e. it includes macropores, transitional pores, and a portion of the
micropores, i.e. those penetrable by the n-hexane molecule. The specific volume
in n-hexane of the coals described in Fig. 5.3 and, in fact, of all but one of
the coals examined declined with temperature from about 350
0
e on indicating con-
solidation. The decline in this volume is much steeper for the softening coal
(Fig. 5.3b), evidently due to melting at about 350
0
e. The specific volume in
mercury for the nonsoftening coal declines monotonically, indicating a shrinkage
of the whole particle. In contrast, the specific volume of the softening coal
goes through a maximum at about 500
0
e indicating mild swelling due to bubble
formation, followed by sharp shrinkage signalling the completion of rapid devola-
tilization and the resolidification of the particles.
An interesting comparison is provided in Figs. 5.4 a, b from the same work of
Toda comparing the volume difference V
Hg
- V h with the total volume of pores
o n- ex
above 150 A as determined by mercury penetration. The close agreement between
these two volumes at all pyrolysis temperatures clearly implies that those coals
did not possess significant pore volume with openings between a size penetrable
o 0
by n-hexane ( ~ 8 A) and 150 A. Moreover, no such volume is produced during
pyrolysis. The absence of pores with openings between about 8 ~ and 150 ~ is
certainly not a universal property of coals (see table 5.1). Figure 5.4 a, b
also shows that the volume V
H
- V h for the softening coal passes through a
g n- ex
maximum coincident with bubble formation. This volume, which belongs to pores
o
of size 150 A or higher, subsequently declines with the disappearance of the
85
0 . 9 ~ - - - - - - - - - - - - - ,
a
800 400
0.5 L..----L_--L_--1..._....L.._-'--_.l----J
o
01 0.8
.....
...
E
u
CI>
E
:l
0.7
0
>
u
-
u
CI>
C.
Cf)
0.6
Pyrol ysi s temperature (Oe)
Fig. 5.3. Specific volumes in mercury and n-hexane of a nonsoftening (a) and a
softening (b) coal as a function of final pyrolysis temperature at heating rate
3
0
C/min (source: ref. 121).
1200 800
b
o 400
Pyrolysis temperature (Oe)
0.12
a
01
.....
...
E 0.08
u
CI>
E
:l
0
0.04
>
CI>
...
0
a..

a
e

0
0 800 1200
Fig. 5.4. The volume difference V
Hg
- V
n
-
hex
(I) and the volume obtained by
mercury penetration (0) for a nonsoftening coal (a) and a softening coal (b) vs.
final pyrolysis temperature at heating rate 3
0
C/min (source: ref. 121).
86
bubble structure and sustains no further change above 600
0
e. For the non-softening
coal, the volume V
H9
- V
n
-
hex
shows no significant change with the maximum pyrol-
ysis temperature.
To characterize the evolution of the microporous structure with heat treatment,
Tocta measured specific volumes in helium and methanol as well as in n-hexane.
Figure 5.5 a, b shows these volumes as a function of the final pyrolysis tempera-
ture for the non-softening and the softening coals examined in the earlier
figures 5.3, 5.4. For both coals the three specific volumes show a rapid decrease
after about 400
0
e suggesting drastic changes in the microporous structure. For
the nonsoftening coal, the volumes in methanol and helium reach a minimum at
800 - gOOOe above which they increase again. For the softening coal, all volumes
1200 800 400
0.5L-_..L-_..L-_...L-_...L-_-'-_-'-_....I
o
0.9
a b
0.8
go
"
...
E
u
IU
....-n-hexane ......... n-hexane
E
:::J
0.7
0
>
. ~
-
u
IU
0.
U)
0.6
methanol
Fig. 5.5. Specific volumes in n-hexane, helium and methanol of a nonsoftening
(a) and a softening (b) coal vs. final pyrolysis temperature at heating rates
3
0
e/min (source: ref. 122)0
continue to decline up to the highest pyrolysis temperature utilized (1200
0
e).
The difference between the two types of coal was attributed to the susceptibility
of softening coals to alignment and consolidation of their crystallites. In
nonsoftening coals, alignment and consolidation are negligible, the primary
volume changes being due to widening or narrowing of apertures. Initially,
apertures are enlarged due to volatile removal leading to a decrease in the
87
helium and methanol volumes. Beyond 800 or 900
0
C, apertures start being sealed
by cross-linking reactions leading to an increase in the two specific volumes.
Methanol is known to penetrate very fine pores and cause a certain degree
of swelling. Because of this penetration, or inbibition, the volume in methanol
is lower than that in helium (at least up to 800
0
C) despite the reverse order in
their molecular size. The difference V
n
-
hex
- V
MeOH
was chosen as a measure of
the volume of micropores with openings smaller than those penetrated by n-hexane,
o
i.e. smaller than about 8 A. Figure 5.6 a, b plots the difference V
n
-
hex
- VCH30H
versus the final pyrolysis temperature for the two coals. Both coals display a
maximum, the nonsoftening coal at about 700
0
C, the softening coal at about 600
0
C.
The increasing part of the curves is mainly due to volatile evolution while the
0.1 2 r------------,
a
C'
.....
...
0.04
E
u
:I:
0

~
>
0.12
I
>C

.I:
I
0.08 c:
>
0.04
Fig. 5.6. The volume difference V
n
-
hex
- V
MeOH
vs. final pyrolysis temperature for a softenlng
(a) and a nonsoftening (b) coal (source: ref. 122).
88
decreasing part mainly due to sealing of pore apertures by crosslinking reactions.
An additional feature of the softening coal is the minimum at about 400
0
e, obvi-
ously due to loss in pore volume caused by melting.
The results just discussed were obtained with very low heating rates (3
0
e/min).
The differences between softening and nonsoftening coals are expected to be
more pronounced at higher heating rates which accentuate softening and swelling
properties, Unfortunately, very little work has been conducted on the changes
in the porous structure of softening coals under conditions of rapid pyrolysis.
Figure 5.7 (a) shows the changes in the size distribution of transitional pores
of a softening coal heated to 500
0
e at heating rates about 200
0
e/s. The main

change is the elimination of pores in the range 15 to 60 A. The total pore

volume, however, increased indicating increase in the macropore range due to
-- Row Coal (N2 pore IIOI::Q,032cm
3
/g)
---- After pyrolysis (N2 porevol.:::Q,043 cm
3
/g)
-- Raw cool (N
2
pore 1101::0.014 cm
3
/g)
---- After pyrolysis (N
2
pore 1101,::0.024 cm
3
jg)
Pore diameter
0.03 0.02
r----'
I I
___J L ,
I
I
I
a
b
r--------
I ________ J
0.01
10'
E
:l-
10
'" "-
'"
E
<>
0
<
"-
10-'
:::
10-
2
0
10'
E
:l-
10
'" "-
'"
E
<>
0
r
"-

>
.-:
10-
0.01 0.02 0.03
Pore diameter
Fig. 5.7. Pore volume distribution of a hvc bituminous coal (a)
and a subbiturninous coal (b) before and after pyrolysis at 500
0
e
for 30 s (source: ref. 62).
89
bubble formation. At higher temperatures and heating rates, the rapid evolution
of volatiles leaves behind large voids occupying almost t h ~ entire volume of the
swollen particle with the solid matter forming a lacey network of thin shells.
A single void occupying almost the whole particle is called a "cenosphere".
Photographic studies of these phenomena (refs. 123, 124) have documented the
geometry of the large voids but have not given information about changes in the
pore volume distribution at the lower end of the size scale.
5.2 THE EFFECTS OF PRESSURE AND PARTICLE SIZE ON PRODUCT YIELDS
5.2.1 The effect of pressure
From the limited data available on the p r ~ s s u r e rlependence of the weight ;oss
we reproduce here Figs. 5.8 and 5.9 from the work )f Anthony et al. (refs. 125,
126). Figure 5.8 shows the weight loss of a hva bituminous coal (Pittsburgh No.8)
as a function of temperature for two pressure levels. Above about 600C the
weight loss at 69atmis substantially lower than that at atmospheric pressure.
Figure 5.9 shows the weight loss at lOOOC as a function of pressure for the same
high volatile bituminous coal. Even at atmospheric pressure, the weight loss is
substantially below its vacuum value. The difference, however, should be less
pronounced at lower temperatures.
1200
H
2
.690tm
o He. 69 otm
A He, lotm
t;,. N
2
lotm
70
60
50
-
~
0
.......
II)
40
II)
0
- .s::
30 01
CI)
~
0
20
/"
10
0
200
Fig. 5.8. Weight loss vs. temperature at two pressure levels
for a hva bituminous coal (source: ref. 125).
90
60
50
t
10-
1
1.0
Pressure (aIm)
10 100 10'
Fig. 5.9. Weight loss vs. pressure at 1000
0
C for pyrolysis
and hydropyrolysis of the Pittsburgh No.8 coal (source:
ref. 125).
30
0
25
0
~
I aim He
20
0
J:: .69 aim He
0'
OJ
Jt
15
"0

OJ
10
~
0
I-
5
0
a
200 400 600 800 1000
Peak temperature
(Oe)
Fig. 5.10. Tar yield vs. temperature at two pressure
1eve1s for the Pi ttsburgh No. 8 coal (source: ref. 63).
91
An investigation of the pressure dependence of individual product yields for
the Pittsburgh No.8 bituminous coal was conducted bySuuberg (ref. 63). Figures
5.10, 5.11 summarize some of his results. The tar yield is shown in Figure 5.10
a function of temperature for two pressure levels, 1 and 69 atm. The yields at
the two'pressures begin to diverge at about 700
0
e. At 10000e the yield at 69 atm
is almost half its atmospheric value. Figure 5.11 plots the yields of various
classes of products versus pressure at 10000e. The basic trend is very clear.
As the pressure increases the yield of tar decreases while the yields of hydro-
carbon gases increase. Since tar is the predominant product on a weight basis
its decrease outweighs the increase in the gases, whence the decrease in the total
volatiles (weight loss).
I : J ~
o
"0
o
0: 6
1 ~
10- 5 lQ-3 iO I 10
1
Pressure (atm)
Fig. 5.11. Product yields vs. pressure for the pyrolysis of
the Pittsburgh No.8 coal in helium at 1000
0
e (source: ref. 63).
In a study of a hvc bituminous coal Gavalas and Wilks (ref. 62) found a sig-
nificant pressure dependence of product yields at temperatures as low as 500
0
e.
The pyrolysis tar of the Kentucky No.9 was separated by gel permeation chromatog-
raphy into three molecular weight fractions and it was observed that the vacuum
tar had larger molecular weights than the atmospheric tar (ref. 7.). The pressure
dependence of the product yields in the pyrolysis of a lignite was reported by
92
Suuberg et al. (refs. 63,64). Below 700C the yields were pressure independent but
above that temperature the yield of gases increased with pressure as shown in Fig.
5.12 for the case of methane. In the same figure the yield of other hydrocarbons
and tar is observed to decrease with pressure due to the decline of the tar com-
ponent. The total yield of volatiles, not shown in the figure shows a modest de-
cline with pressure.
2.5
-----
~ 1.5
'"
>-
:J!
2 1.0
Q;
::;
0.5
Do 69 atm He
o I atm He
x 10-
4
atm He
00 0
o
x ~ x
" _ ~
1100
Peak temperature ICC)
Fig. 12. Methane yield vs. temperature at three pressure
levels for the pyrolysis of a lignite (source: ref. 64).
5.2.2 The effect of particle size
Very few experimental data are available concerning the particle size dependence
of product yields. Figure 5.13 describing the pyrolysis of a bituminous coal shows
that increasing the particle diameter by a factor of ten causes only a modest de-
crease in the weight loss and the yield of tar and an increase in the yield of
gaseous products. A similar effect was found by Anthony et al. (ref. 126) for the
Pittsburgh bituminous coal at atmospheric pressure and 10000C. Gavalas and Wilks
(ref. 62) and Solomon (ref. 70) also observed small effects of partlcle size in
the pyrolysis of high volatile bituminous coals in the pressure range vacuum to
2 at and at temperatures 500-100
0
C.
Significant size effects were observed in the pyrolysis of a subbituminous coal
(ref. 62) as shown in Fig. 5.14. The tar yield shows a mixed trend probably due
to difficulties in its quantitative recovery. The yield of gases, however, shows
a substantial increase with particle size.
93
-
-
-
40 -
70...----,.---,-.---,-,-----,
H
2
.690tm
o He. lotm
30 I I I
o 300 600 900 1200
Mean particle diameter (fJm)
Fig. 5.13. Weight loss vs. particle size for the ptrolysis and
hydropyrolysis of the Pittsburgh No.8 coal at 1000 C at 1 atm
He. (Source: ref. 125).
In studying particle size as an experimental variable it is necessary to take
some precautions to avoid simultaneous variation of chemical composition due to
maceral enrichment. If a quantity of coal is ground and sieved, the size fractions
will not be uniform in maceral content (and hence chemical composition) because
of the different grindability properties of the macerals. Moreover, the small
size fraction usually becomes enriched in minerals, especially pyritec The diff-
erentiation in maceral content can be avoided by a simple modification of the
grinding-sieving procedure. The whole coal is ground and a particular sieve
fraction separated representing the largest size desired. A portion of this
fraction is ground again and a second and smaller sieve fraction is separated, etc.
5.3 ANALYSIS OF HEAT TRANSFER
In a broad range of experimental conditions heat transfer is sufficiently rapid
compared to chemical reactions so that it does not influence the weight loss and
yield of individual products. In the first subsection we derive a criterion for
the absence of heat transfer limitations. In the second subsection we briefly
discuss some attempts towards a theoretical analysis of simultaneous heat trans-
fer and kinetics.
5.3.1 Criteria for the absence of heat transfer limitations
To estimate the magnitude of heat transfer effects in coal pyrolysis we shall
examine a very simple model problem that possesses the main features of the real
94
28
[; 450 ~ m
o 110 IJrn
...... ~
o
~
o
10
U
I
U
20\.- --'- ---'
8
[; [;----
6 ------------ [;
[; ..31----_
-'--- 0 ---8
6-
4'-- -'- ---'
~ 44
~ 6
N ~
8 40 0 [;
I o - - - - - - ~ - - - - - - ~
o
u
Helium pressure (N/m
2
i
Fig. 5.14. Product yields as percent of weight loss vs. pressure
for two particle sizes in the pyrolysis of a subbituminous coal
(source: ref. 62).
problem. In the model problem we consider a solid sphere with density, heat capac-
ity and thermal conductivity p , c , A immersed in a hot gas with bulk tempera-
s ps s
ture Too' The heat transfer between gas and sphere is described by a heat transfer
coefficient h. The interior of the sphere is cooled by a constant and spatially
uniform volumetric heat sink of strength bT simulating the heat effect of the
pyrolysis reactions. The equations governing this simple situation are
1 8
a --
s 2
X 8x
8T
+ at
(5.1 )
x = 0: 8T/8x = 0
x = a: A
s
8T/8X = h(Too-T)
where as = AS/PSCps and a is the particle radius.
Equations (5.1)-(5.3) can be written in the dimensionless form
(5.2)
(5.3)
+ 13(1-8) = l
8T
o
(5.4)
(5.5)
E; = 1
a8
: = -A8
95
(5.6)
2 2
where E; = x/a,T = astla , S = a biAs' y = halAs' 8
The solution of this problem can be written as
8 = 8
ss
+ 8
us
where
!-;:; .k:
_ 1. exp(S2E;)-exp(-S2E;)
E;
(5.7)
(5.8)
00
8
us
'= L gn
exp
n=l
[
2 ]sinl:;E;
-(I:;n + S)T ---E;--n-- (5.9)
where gn depend on the initial particle temperature and 1:;1 < 1:;2 < are the
positive roots of the equation
tan I:; = _1:;_
1-y
As T+oo,8 tends to the steady state profile 8
ss
for which
k
8
ss
(0) = 1 _ 2yS2
1: h 10: h h
S2[exp( 62)+exp( -6
2
)]_ (l-y) [exp( 62)-exp( -6 2)]
(5.10)
(5.11)
The preceding expressions can be used to investigate the time required to
reach the steady state and the shape of the steady state. The length of time is
2
determined by the quantity I:; + 6, the smallest coefficient of T in the exponen-
1
tials of Eq. (5.9). This time can be defined by
T
1
1
--2
S+I:;
1
(5.12)
showing that the presence of an endothermic reaction speeds up the approach to
2
steady state. The quantity I:; is a function of y(eq. 5.10) which can be
1
estimated as follows
ha A
y = - = Nu .-Jl
\ AS
The Nusselt number is usually larger than 2 while Ag/As 0.1-0.5. Hence y>0.2.
An examination of (5.10) reveals that y>0.2 implies I:; (y) >I:; (0.2)=0.593, hence
1 1
T <
1 6+0.593
(5.13)
or in terms of the dimensional time
2
t < __
1 6+0.59 as
96
2
When the heat effect due to reaction is insignificant, t
j
< 1.69 a/as'
We now examine the steady state profile. The quantity 8 (0) = (T -T(O))/T
ss 00 00
represents the maximum difference between particle temperature and gas tempera-
ture. Eq. (5.11) shows that as 8
ss
as it should. The net effect of
the pyrolysis reactions can be endothermic (b>O) or exothermic (b<O). In a sub-
stantial range of experimental conditions 61 and 8
ss
(0) can be approximated by
8
ss
(0) = < oe 2.56
We take as an example a reaction temperature of 1000oK, a heat of reaction 30cal/g
3 3
and ps=l g/cm , As=lO cal/cms K, a=50 To obtain b we need the rate of de-
volatilization. At 10000K we may assume a devolatilization rate of 3 g/gs so
3 2 3 3
that b = 30x3/1000 = 0.09 cal/cm s Kand 6 = 0.005 xO.09/10 = 2.25xlO Under
_3
these conditions the approximate expression above holds, i. e. 8ss(0)oe5.6xlO or
Absence of heat transfer limitations requires that the time to reach steady
state is short, compared to the reaction time, and that the steady state tempera-
ture difference is small:
2
__ t
as 6+0.59 r
2.5 61
These are equivalent to
6 0.4
(5.14 )
(5.15 )
(5.16 )
(5.17)
The experimental reaction time has the order of magnitude l/r
o
where r
o
is the
rate of devolatilization in g/gso Substituting the experimental time for t
r
in
Eq. (5.14) and using dimensional variables in (5.15) we obtain
0.59
The last equation can be rewritten as
(5.18 )
(5.19 )
(5.20)
97
The ratio 11IHI/Tc
p
fs always much smaller than unity, hence (5.18) implies (5.19).
Our final conclusion then is that Eq. (5.18) provides a necessary and sufficient
criterion for the absence of heat transfer limitationsc It should be kept in mind
that r
o
is the experimentally measured rate and when Eq. (5.18) is violated, the
reaction 'rate r
o
already includes the retarding effects of heat transfer.
As an illustration of the above criteria we may take the rate expression r =
o
870 exp(-13,200/RT) obtained by Suuberg et al.(ref. 71) for the devolatilization
of a bituminous coal in the range 1000-2l00
o
K. Curve A in Fig. 5.15 is a plot of
the equation
2
a ( 13200)
a 870 exp ,- ~ = 0.059
s
which defines on the T-a plane the region free of heat transfer limitations.
The above simple analysis can be supplemented to take into account two effects,
the radial convective flow in the gas film surrounding the particle, and the swell-
ing of the particle.
The effect of the convective flow in the film can be taken into account (ref. 127)
by changing boundary condition (5.3) to
x=a: \3T/3X
When Gc /h
pg
we have
Gc (T - T)
pg 00
(5.21 )
Using G=r p (4TIa
3
/3)j4TIa
2
,h=NuA /a
o s 9
(5.22)
(5.23)
Typical values for the first two factors are :\Cp/AgC
pg
'" 6, Nu=2, hence
2
GC
pg
'" r oa
h O:s
It is now seen that Eq. (5.18) implies (GC
pg
/h)l and the boundary condition
(5.21) reduces to the simpler (5.3).
A more serious complication is encountered in the thermal analysis of strongly
swelling coals. Bubble formation in the molten coal increases particle size and
decreases density and thermal conductivity. Since the decrease in density out-
weighs the decrease in conductivity, the thermal diffusivity experiences a modest
decrease. The predominant effect, however, is due to the increased particle size.
If the swelling factor is known then the heat transfer analysis should be based
on the expanded particle rather than the initial particle. The situation, how-
ever, is more complex because the swelling factor itself depends on temperature
and initial particle size.
98
28o,------,----,------,----,__--,
240
200
160
-0
"
'"
120
o
(L
80
40
No limitations
Transfer limitations
/Film moss transfer
Heat transfer
o
T (OK)
Fig. 5.15. The region of heat transfer (A) and film mass transfer (8)
limitations for the pyrolysis rate r =870 exp(-13,200/RT)g/gs and
o
a
s
=2.2xlO-
3
cm
2
/s.
5.3.2. Analysis of combined heat transfer and kinetics
Among the few modeling studies of coal pyrolysis including the effects of heat
and mass transfer are those of Mills et al. (ref. 127) and Jones and Mills (ref.
128). The kinetics are described by an empirical set of reactions quite similar
to that used in other studies,

k2
metaplast-
metaplast
semi coke + volatiles
The transport aspects of the model are treated in a rather detailed fashion, es-
pecially the energy equation which incorporates convective terms due to particle
swelling. The swelling is described in an empirical and not altogether satisfac-
tory way, therefore the specific numerical results obtained are not very reliable.
Nevertheless, they point out, at least qualitatively, the gross effects of heat
transfer limitations. A sample of the numerical results of Mills et al. (ref.
127) is given on Figure 5.17 showing the cumulative evolution of total volatiles
as a function of time for different particle sizes. The ultimate weight loss is
predicted to be independent of particle size in view of the lack of secondary
99
reactions in the above kinetic model. Hence, the particle size is predicted
to influence only the time required to complete
5.4 OF MASS TRANSFER
Mass transfer effects are experimentally manifested in the dependence of product
yields on pressure and particle size reviewed in section 5.2. The qualitative in-
terpretation of these effects is straightforward. Amolecule generated in the
interior of the coal particle requires a certain length of time before it is re-
moved from the particle. During this time the tar molecule may become reattached
to the condensed phase to eventually decompose into light gases and solid char.
Slow mass transfer thus increases the extent of secondary reactions causing a
higher yield of gases, a lower yield of tar and a lower overall production of
volatiles.
Mass transfer includes two processes, intra-particle and film mass transfer.
Intraparticle mass transfer involves the transfer of a product molecule from its
generation site in the condensed phase to the external surface of the particle.
Film mass transfer involves the transfer of the molecule from the external part-
icle surface to the bulk of the gaso
-
40

0
-
30
If)
If)

20
-
.r::.
0'
10
CD
3:
a
0.001 0.01
0.\ 10
Time
(s)
Fig. 5.16. Model calculations of weight loss vs. pyrolysis time for
several particle sizes ref. 127).
Film transfer proceeds by the same mechanism in softening and nonsoftening
coals although in the former case one might have to consider the change in particle
100
size due to swelling. The mechanism of intraparticle mass transfer, on the
other hand, differs profoundly between these two types of coals. In softening
coals the transfer of products occurs by nucleation, growth and coalescence of
bubbles during the plastic state of the particles. In nonsoftening coals diffusion
and forced flow through the pore structure constitutes the transfer mechanism.
In the first subsection we examine film transfer seeking criteria for the absence
of transfer limitations. In the second and third subsections we consider intra-
particle mass transfer in softening and nonsoftening coals seeking a semiquanti-
tative interpretation of the effects of pressure and particle size.
5.4.1 Film mass transfer
To develop simple criteria for the absence of transfer limitations we consider
a spherical particle immersed in a radially uniform flow field generated by the
flux of productso The products are lumped into two species: gas and tar, the gas
including all low molecular weight products and the tar including all products
with molecular weight above one hundred. The lumping is employed so that we may
use the equations of binary diffusion and thus obtain analytical results. Binary
diffusion at steady state (pseudo-steady state) is described by
(5.24)
(5.25)
(5.26)
where it is assumed that the only convective term is in the radial direction. In
Eqs. (5.24) and (5.25) NT,NG'YT'YG are molar fluxes and mole fractions for the
tar and the gas, c is total concentration and D
TG
is the binary diffusion coeffic-
ient. These two equations are equivalent (as shown by addition), therefore only
one need be considered.
2
In the absence of any source terms in the gas phase the quantities FG=x NT'
2
FG=X N
G
are constant and equation (5.24) can be integrated to
YT(a)=vT(l-e-o)+e-OYToo
where YT(a)'YT are the mole fractions of tar at x=a and x=oo, vT=FT/(FT+F
G
) and
00
(5.27)
Equation (5.26) can be used to analyze two related but distinct problems. In
the first problem YT(a) and F
G
are considered known and Eq. (5.26) is used to de-
termine FT' Since v
T
and 0 are functions of F
T
the determination of F
T
is implicit.
In a broad range of conditions 01 (see below) and Eq. (26) provides the
101
approximate solution
which after setting c=p/RT, YT=PT/P can be rewritten as
D
TG
F
T
=~ (PT(a)-PToo)
where R is the gas constant.
(5.28)
(5.29)
Up to this point the analysis is similar to that of Unger and Suuberg (ref. 129).
These authors assume that Pr(a) is an equilibrium pressure corresponding to the
concentration of dissolved tar, or metaplast, which is uniform throughout the par-
ticle. The latter assumption probably does not hold in most conditions of inter-
est and will not be adopted in the following.
The effects of pressure and particle size can be explored to some extent with-
out making any statement about the value of PT(a) which largely depends on intra-
particle kinetic and transport processes. Expressing the molar rate of tar prod-
uction per unit mass of coal we obtain
(5.30)
(5.31)
where Pc is the density of the coal particleo The rate of tar evolution appears
inversely proportional to a
2
, but this does not take into account the dependence
of PT(a) on particle size. The effect of pressure is twofold. The diffusion co-
efficient is inversely proportional to pressure while PToo is approximately pro-
portional to pressure. Thus as the pressure increases the rate of tar generation
decreases resulting in a higher "metaplast" concentration in the coal melt and a
smaller cumulative tar yield.
Eq. (5.30) is coupled through the term PT(a) to the kinetics governing the
evolution of the metaplast concentration in the coal meltc Some results can be
obtained, however, without kinetic considerations. In experiments carried out by
the captive sample technique, PT = a because of tar condensation on the cold reactor
walls, therefore
3 DTG(po)p
o
PT(a)
r
T
= p RT paz-
c
where p =1 atm. This expression implies that the ultimate tar yield depends on the
o 2
product a p rather than on p and a independently. The scarce data available on
particle size dependence do not allow evaluation of this particular result. As
we shall see below, however, film mass transfer is in most cases very fast, there-
fore the pressure and particle size dependence of the tar yield must be mainly
102
due to intraparticle mass transfer processes.
We now return to Eq. (5.26) to derive a criterion for the absence of film mass
trans fer 1imitations. Such a criteri on may be expressed in the form
which can be rewritten with the help of Eq. (5.26) as
(5.33)
Clearly, this criterion is satisfied when 01. The parameter 0 is a kind of a
Peclet number depending on particle size and temperature but independent of pres-
sure since DTGcrl/P.
To estimate 0 we relate v
T
and F
i
to the experimental rates of tar and gas prod-
uction:
i=T,G
(5.34)
(5.35)
where M
T
and M
G
are the mean molecular weights of tar and gases, about 300 and 30
respectively. Introducing (5.35) into the definition of 0, Eq. (5.27), we obtain
The parameter 0 may now be estimated by using
2 3
cm /s,T
l
=800 K,pc=1.2 g/cm , rT/MT+rG/MG=ro/30.
=870 exp(-13,200/RT) we obtain
(5.36)
I. 75
the estimates atm
Using the previous expression r
o
1.14 a
0- --'-"--'-'---"-.
O. 75
T
(
13,200)
exp -
(5.37)
with a in and T in oK. To obtain concrete results, we arbitrarily set 0=0.1
as the limit of transfer limitations. The value 0=0.1 defines curve B in Figure
5.16. This curve lies remarkably close to curve A defining the region of heat
transfer limitations. The location of both curves is specific to the rate ex-
pression chosen to represent the rate of pyrolysis.
5.4.2. Intraparticle mass transfer in softening coals
We are here primarily concerned with the evolution of tar and gases during the
plastic state of coal. In this situation mass transfer consists of two
in series: diffusion through the molten coal to some internal surface, that of a
103
bubble or a pore; and transport with the bubble or through the pore to the surface
of the particle. The role of preexisting pores is not well understood. As dis-
cussed earlier in connection with Fig. 5.7, a certain fraction of preexisting
pores 60 A) collapse during pyrolysis perhaps due to surface tension effects.
Pores in the range 60-300 Awere preserved but in this case one could not distin-
guish between preexisting pores and pores generated by the evolution of bubbles.
It appears likely that the major part of mass transfer occurs via bubbles while
preexisting pores playa relatively minor role.
A detailed theoretical analysis of mass transfer based on the nucleation,
growth and coalescence of bubbles was carried out by Lewellen (ref. 130). The rate
of bubble nucleation employed in this analysis depended on several variables
including the rate of pyrolysis and contained some adjustable parameters. Bubble
growth was described by the Navier-Stokes equations assuming an idealized geom-
etry. Coalescence and bursting were postulated to occur when the surfaces of
growing bubbles meet one another or reach the external particle surface. The
bubble transport equations were interfaced with the chemistry by assuming that
the volatiles experience secondary reactions on the bubble surface. The calcula-
tion of extent of these secondary reactions employed bubble surface area and
pressure which, in turn, depended in a rather complicated fashion on particle
size and external pressure. The analysis thus established the dependency of
the volatile yield on particle size and external pressure. The predicted
dependence on particle size was in qualitative agreement with experimental data
but the predicted dependence on pressure was not satisfactory.
Although adjustments can be made to improve the predicted dependence on ex-
ternal pressure, the coupling between chemical reactions and bubble dynamics
remains questionable. As formulated, the analysis does not distinguish between
tar and gases and does not take into account considerations of phase equilibrium
between bubbles and coal melt which as we shall see below hold the key to the
evolution of tar vapors. In spite of these deficiencies and its complexity,
Lewellen's analysis contains many useful ideas especially with regard to bubble
growth.
To take into account the constraints imposed by thermodynamics we must disting-
uish between gases and tar vapors. Liqht hydrocarbon gases, water vapor and
carbon oxides have very low solubility in the coal melt at the reaction temperature,
therefore they nucleate quite rapidly to bubbles which grow, coalesce and finally
burst through the boundary of the molten particle. Tar molecules on the other hand
possess relatively low vapor pressure, therefore make a negligible contribution
to the nucleation rate. Nevertheless, they do exert a vapor pressure in the bubbles
at equilibrium with their concentration in the coal melt adjacent to the bubble
interface. It appears very likely that secondary reactions involve primarily
the tar dissolved in the coal melt and to a lesser extent the tar vapors in the
104
bubbles or outside the particle.
The partial pressure of tar in the bubbles is at equilibrium with the coal
melt at the bubble surface and not necessarily with the bulk of the coal melt.
Whether or not substantial gradients of tar concentration exist between the bulk
and the interface can be determined, in principle, by comparing the characteristic
time for diffusion in the melt with the time for chemical reactions. If D
TM
is
the diffusion coefficient of tar molecules and x
M
is the corresponding diffusion
path, the characteristic diffusion time may be defined as
The diffusion coefficient can be crudely estimated from the Stokes-Einstein
equation
where w is the viscosity of the melt and a the radius of the diffusing molecule.
The viscosity of the coal melt is not well defined because of the transient nature
of the plastic stage and the presence of bubbles and solid particles. For esti-
6
mation purposes we may use values in the range l-SxlO g/cms (refs. 112,131). Such
values correspond to low heating rates (a few C/min) and temperatures about
420C. The transient viscosity at higher heating rates and temperatures could be
6 0 0
considerably lower. If we use w=3xlO g/cms and a=8 A,2.S A for tar and gases
we obtain the estimates DTM=2X10-14cm2/s, DGM=SxlO-14cm2/s at SOOC. These values
are lower by many orders of magnitude than those employed by Attar (ref. 132) and
Unger and Suuberg (ref. 129) in their analysis of diffusion in the coal melt.
The diffusion path x
M
is the thickness of melt layers between adjacent bubble
surfaces and depends on the rates of bubble nucleation, growth and disappearance.
Bubble nucleation in coal melts has been analyzed by Attar (ref. 132) but his
estimates of nucleation rates are probably too high because of overestimating the
diffusion coefficient. Bubble growth, coalescence and disappearance have been
analyzed in the aforementioned work of Lewellen (ref. 130). The lack of reliable
physical properties and the difficulty of describing bubble coalescence constitute
a formidable problem in estimating x
M
. However, we can inquire about the values
of x
M
for which the diffusional resistance in the melt becomes important. For a
pyrolysis time of lOs (at diffusion starts being important when tMS=ls
which corresponds to xM=lO A, if we use the previously estimated values of D
TM
.
Even if the viscosity were lower by a factor of one hundred, tar transport from
the bulk of the melt to the bubble surface would remain one of the rate determining
processes. The above observations suggest a qualitative explanation of the role
of external pressure in determining the yield of tar. Increasing the external
pressure causes the bubble size and the rate of bubble coalescence and bursting
through the particle surface to decrease. As a result of the longer bubble
105
residence time and tne higher bubble pressure,the partial pressure of the tar
increases and tar evolution is slowed down. The result is more extensive
secondary reactions and -lower cumulative yield of tar.
The tar yield also depends on particle size. Larger particles require larger
internal bubble pressure to overcome the viscous resistance to bubble growth. As
before, larger internal pressure adversely affects the tar yield. Howevr, the
particle size effect has not been adequately documented and seems smaller than
the pressure effect.
5.4.3. mass transfer in nonsoftenino coals
Mass transfer in these coals proceeds through the porous which
relatively stable during pyrolysis. Much of the rilethodology Of mass tl'ansier in
porous solids has been developed in the context of heterogeneous catalysis and is
not directly applicable to coal and coal char. The latter materials possess a
very broad pore size distribution including micropores 0.0004-0.0012 trans-
itional pores 0.0012-0.03 and macropores 0.03-1 diameter. This broad size
distribution poses two theoretical difficulties: (i) diffusion in the micropores
is activated and suitable diffusion coefficients have not been measured at pyrol-
ysis temperatures (ii) coal particles with size of the order of 100 typical
of many experimental situations, cannot be strictly treated as continua using
an effective diffusion coefficient.
The microporous space of coal and char has molecular sieve characteristics.
Diffusion through this space depends on molecular size and is very slow and acti-
vated. The pyrolysis products must first slowly diffuse through a domain pene-
trated by micropores only until they reach the surface of some transitional pore
or macropore and then be transported by convection and diffusion to the surround-
ing gas. The coal or char particle may then be considered as a two ohase region,
the coal phase, including the micropores, and the void phase consisting of pores
larger than 12 Ain diameter. The average diffusion path that a molecule in the
coal phase must transverse before the void phase depends on the pore size
distribution in a complex fashion which has not yet been investigated. However,
for commonly encountered materials some tentative estimates place it in the range
500-2500 A. The diffusion coefficient of molecules such as CH
4
, H
2
0, CO etco in
the coal phase is probably larger than 10-
10
cm
2
/s so that under most conditions
such molecules require about 1 s before they reach the void phase. Given the low
reactivity of these molecules such residence time is too short for any secondary
reactions.
Because of their size, molecules in the tar range diffuse much more slowly in
the coal phase. Although the diffusion coefficients are not known they could be
smaller than 10-
12
cm
2
/s. The corresponding diffusion times are in the range of
100 s, quite long given the propensity of the tar molecules to secondary reactions.
106
Limitations in the evolution of tar due to this sloW diffusion in the coal phase
are independent of particle size (the diffusion path depends on the pore size
distribution and not on the particle size) and cannot be experimentally separated
from the kinetics of pyrolysis. An empirical device for taking into account the
slow diffusion through the coal phase has been utilized in the pyrolysis model of
Gavalas et al. (refs. 133,134).
Once in the pore space (>12 A in diameter) gases and tar vapors are transported
by the customary mechanisms to the particle surface and from there to the bulk
of the gas. The transport of pyrolysis products through the intraparticle void
or pore space depends on size and external pressure. This is the de-
pendence that has been studied experimentally and constitutes what is nQrmally
called intraparticle mass transfero Mass transfer and kinetics are intimately
coupled and should in principle be analyzed together. However, to combine mass
transfer with an elaborate kinetic model requires the solution of a coupled system
of several stiff partial differential equations, a formidable numerical task in-
deed. Russel et al. (ref. 1351) studied th'eoretically mass transfer in combination
with simple empirical kinetics for coal hydropyrolysis. Gavalas and Wilks (ref.
62) obtained some qualitative results concerning the effects of pressure and par-
ticle size on pyrolysis yields by formulating a simpler and limited in scope prob-
lem. If the instantaneous rates of tar and gas production are known experimentally
an analysis of the mass transfer equations provides the pressure and concentrations
of gases and tar in the pore space. The concentration of tar can then be used as
a qualitative measure of the pressure and particle size effects.
We will now summarize the chief results from reference 62. By assessing the
o
role of pores of various sizes it is found that pores in the range 300-A to 1
constitute the main channels for mass transfer. This pore range is approximated
by a bimodal pore system with radii 0.05 and 0.5 The larger pores are few in
number and hence poorly cross-linked, therefore, transport in the particle cannot
be treated using the customary diffusion coefficient and permeability. Instead,
the following simplified assumptions may be employed. Once a molecule enters the
large pores it is quickly carried to the external surface due to the large trans-
port coefficients in these pores. Thus the large pores effectively shorten the
diffusion path through the smaller pores via their mutual intersections. This
situation is approximately described by defining a smaller effective particle
size and considering diffusion through the small pores alone. Binary effective
diffusion coefficients can then be employed.
The multicomponent mass transfer problem can be further simplified by lumping
the various gases into two or three groups according to molecular size. When
pyrolysis is carried out in hydrogen or helium, three components are required:
hydrogen (helium), gases (C0
2
,H
2
0,CH
4
etc.) and tar. When pyrolysis is carried
107
out in the atmosphere of pyrolysis or combustion gases, or in nitrogen carrier,
two components are sufficient, gases and tar. Although reference 62 treats both
the ternary and the binary problem we will here give the results for the binary
problem only.
Neglecting accumulation terms, the mass balance equations for gases and tar
in a spherical coal particle are given by
G,T
(5.38)
where N
i
are the molar fluxes and Yi the molar rates of production per unit vol-
ume. The source terms Yi are assumed spatially uniform and known, from the meas-
urements. In a more complete analysis, of course, Yi would have to be related to
the concentrations via the kinetics and would thus vary with the radial position.
With Yi treated as independent of r, Eq. (5.38) is immediately integrated to
r
N
i
= Yi 3"
i = G,T (5.39)
The quantities N
i
can be described by the approximate flux model
N. = -c .. Q2.
d
d + N ~
1 llJr 1
i = G,T (5.40)
where S , ~ are permeability of the coal and viscosity of the gaseous mixture while
c
i
are molar concentrations and p is the pressure. In Eg. (5.40) the flux consists
of two terms, one due to a pressure gradient and the second, Ni, due to diffusion.
The diffusive fluxes Ni are given by
dC
G
=
N ~
N.yXG-NGx
T
- ~ +
dr
D
GK
D
GT
dC
T
N ~
NGxrN.yx
G
_.J:. +
F=
DrG
D
GT
where
D
i
k'
D
GT
are the
given by
D
ik
s *
=3" D
ik
(5.41 )
(5.42)
effective Knudsen and binary bulk diffusion coefficients
G,T
*
where D
ik
is the Knudsen diffusion coefficient in a single capillary of the type
considered, D ~ i is the bulk diffusion coefficient at atmospheric pressure and
108
P'Pat are pressure and atmospheric pressure. The pressure p is a function of
location within the particle.
Combining Eqs. (5.39)-(5.42) we obtain
::' 0 - :: -; [yO D:
K
)- YT (5.43)
::T 0 - :: -; [YT D;K) -Y, ] (5.44)
By adding these two equations and using the ideal gas law cG+cT=p/RT we obtain an
equation for the pressure,
Q2.=
dr
RT(YG+ YT)
D
GK
D
TK
r
(
XX )-3

)l D
GK
D
TK
(5.45)
By inserting into (5.43) and (5.44) the term dp/dr can be eliminated re-
sulting in two coupled equations in cG,c
T
with the initial conditions
i = G,T (5.46)
where Po' x
Go
' x
To
are the known pressure and mole fractions in the bulk of the gas,
Equation (5.45) can be written in dimensionless form as
dW
dE:
Pat 1 + oPT
Po 1+A( VG+PT
V
T
)
(5.47)
E: r/a,V
i
= ci/(po/RT), (i=G,T),W=p/po
*. / *
o YT/YG' P
T
= D
GK
D
TK
B
Although linearly dependent on (5.43) and (5.44), Eq. (5.47) is useful in cer-
tain limiting cases where it can be solved independently. For example, in the
pyrolysis of subbituminous coals the molar production of gases far exceeds that
109
of tar: YTYG and, in addition, xTox
Go
and VTV
G
, so that Eq. (5.47) may be
simplified to
dW
pat 1 + oPT
+ AW
(5.48)
which can be integrated with the initial condition W= 1 at ~ 1 to yield the
maximum dimensionless pressure difference
1+ A{[ W(O)-l = - -A- 1
where
3SPat
A =--
at EjJD*
GK
+ Aat(l + c5PT)B
(1 + A)2
(5.49)
The maximum pressure buildup depends on the dimensionless parameters A, A
at
and B. For any given coal, the parameter A depends on pressure only and repre-
sents the ratio of characteristic times for diffusion and forced flow (due to the
pressure gradient). Large values of A imply that forced flow is the predominant
mass transfer mechanism while small values of A imply that diffusion is the main
mode ,of mass transfer. The parameter A
at
is simply the value of A when P=Pat.
The parameter B is a ratio of characteristic times for diffusion and the pyrolysis
reaction and encompasses the effect of particle size and temperature. For small
values of A (low pressure) transport occurs mainly by diffusion and the intrapar-
ticle concentration profiles of tar and gases are determined solely by B.
To examine Eq. (5.49) in more detail, we note that the permeability S for
straight cylindrical capillaries of radius R and porosity E is given by S = R
2
E/24.
For R = 0.05 jJm it can be seen that Aatl. As long as B is not much larger than
unity (5.49) can be simplified to
l+c5PT A
at
W(O)-l = -2- A(l+A) B
or, in dimensional form
(5.50)
(A/p )(l+A)
o
B (5.51 )
The pressure buildup is proportional to the parameter B which is proportional to
the square of particle size, and inversely proportional to l+A which includes the
effect of pressure. The quantity A/po is independent of pressure. At low pres-
sures Al and the pressure buildup depends on B alone. At high external pressures,
the buildup p(O)-p depends on both B and A and diminishes with increasing pressure.
o
110
Some results from the numerical solution of Eqs. (5.43), (5.44) are presented
in Figure 5.17. The average intraparticle tar concentration is given as a func-
tion of A (dimensionless pressure) for three values of B (function of particle
size and pyrolysis rate). At low pressures the average concentration is a strong
function of particle size but depends very little on pressure. At high pressures
the concentration depends on both pressure and particle size. Inasmuch as the
extent of secondary reactions depends on the average tar concentration, the
above dependence of concentration on pressure and particle size indicates,
qualitatively, the effect of these two variables on the yield of tar. To
describe this dependence quantitatively requires as we have pointed out the joint
analysis of kinetics and mass transfer.
c:
0
-0
...
8=0.0434
-
10-
9
c:
Q)
u
0.0193
c:
0
u
...
0
0.0048
- 10-
10
Q)
0'
0
...
Q)
>

10-
11
10-
4
10-
7
If)
E
u
.......
0
E
10-
8
0'
Fig. 5.17. Model calculations of average
tar concentration, based on pore volume,
as a function of the dimensionless param-
eters A and B (source: ref. 62).
111
Chapter 6
KINETIC MODELS OF COAL PYROLYSIS
In this >chapter we discuss in detail relatively recent models developed in con-
nection with the experimental programs surveyed in Chapters 4 and 5. Earlier model-
ing work, now largely superseded, is altogether omitted.
The models discussed in sections 1 and 2 are phenomenological in nature. They
postulate a set of independent or coupled reactions describing product formation
in terms of various components whose exact nature remains unspecified. The postu-
lated reactions contain a number of parameters which are determined by comparison
with experimental data. In some models, the estimated parameter values fall in a
chemically meaningful range and as a result they relate to some extent to actual
chemical processes. Because of their simplicity, phenomenological models have
proved useful in combustion and gasification, where the total yield and heating
content of products are essential while the chemical structure and amount of indi-
vidual products is not required.
In the last section of the chapter, we will discuss possible approaches to
developing chemical models describing pyrolysis in terms of functional groups and
their elementary reactions. Chemical modeling has not been feasible in the past
because of the complexity of the overall process, the lack of complete grasp on
reaction mechanisms and the lack of sufficiently accurate values for rate param-
eters. Information accumulating from recent spectroscopic investigations of coal
structure and kinetic experiments with model compounds is gradually imoroving
this situation to the extent that chemically based models are becoming reasonable
goals in pyrolysis research. Such models would provide the theoretical framework
for further experimentation and a useful tool in process and development work,
especially in the related areas of hydropyrolysis and liquefaction in which the
chemical structure of products and intermediates is of the essence. The models
discussed in this chapter are purely kinetic. Transport processes are either
ignored altogether or are incorporated within the overall kinetics in a rudimen-
tary and empirical fashion.
6.1 INDEPENDENT FIRST ORDER REACTIONS
Although phenomenolqgical models do not employ actual chemical structures and
elementary reactions, they do attempt to provide chemically reasonable kinetic
expressions by postulating hypothetical species participating in simple stoichi-
ometric reactions. In our discussion below, we will attemot to bring
forth the basic premises of each model as well as its utility and limitations.
112
6.1.1 A single first order reaction
This and the next model describe weight loss (or total volatiles) only, The
simplest possible description is obviously that of a first order reaction
(6.1)
*
W
v
where W
v
is the weight of volatile precursors in a given sample of coal of
*
initial weight W
o
' and W
v
is the initial value of Wv,which is also the ultimate
yield of volatiles,or the ultimate weight loss.
A survey of results using the simple first order kinetics of Eq.(6.1) is
included in the review of Anthony and Howard (ref. 136). Three important
observations made in this survey are particularly noteworthy. First, the
*
ultimate yield W
v
usually exceeds the proximate volatile matter (VM) of coal
obtained by the well defined ASTM procedure: The ASTM procedure entails extensive
secondary reactions inside and on the external surface of the coal particles.
By contrast, the pyrolysis experiments of interest here relate to conditions
restricting secondary reactions. This difference between the extent of secondary
*
reactions provides an adequate explanation for the discrepancy between W
v
and
VM.
The second observation recalled from ref. 136 is that if the rate constant k
is expressed in Arrhenius form,
k = A exp (-E/RT) (6.2)
the values of A and E obtained by comparison with the experimental data exhibit
large variability" In particular, values of E as low as 4 and as high as 45
kcal/gmole have been calculated while the values of A spanned several orders
of magnitude. Some of this variability might be due to the difference in the
type of coal but for the most it results from forcing the experimental data into
a more or less arbitrary kinetic mold. As discussed in ref, 136, when a set
of parallel independent first order reactions is fitted by a single first order
reaction, the calculated activation energy can be smaller than each of the
individual energies.
*
The third observation from ref. 136 is that the ultimate weight loss W
v
often
turns out to be a function of temperature, showing in a clear-cut way the defi-
ciency of treating volatile matter as a pseudospecies, Specific results showing
*
the dependence of W
v
on temperature are included in Figures 4.3-4,5 of the
previous chapter,
In spite of the difficulties noted above, Eq. (6,1) is often used for crude
estimates and comparisons. For example, the
coal and a lignite were fitted by the common
x 10
4
S-I, E = 25 kcalfgmole (ref. 57).
113
weight loss data from a bituminous
*
parameters W
v
fWo = 0.7, A = 6.6
weight loss
matter.
*
Wv(t) = W
v
6.1.2 Several first order reactions for weight loss
A conceptual improvement in the modeling of weight loss was made by Pitt
(ref. 137) who treated coal as a collection of an infinite number of species de-
composing by parallel and independent first order reactions. The rate constants
were assumed to have common A-factor but different activation energies, varying
in a range [E
min
, E
max
] according to a probability density function, f(E). Thus
f(E)dE is the weight fraction of vOI&tile species with constants
having activation energies in [E, E+dE]. isothermal conditions the total
*
is given by (6.3) where W
v
is identified as the proximate volatile
E
max
ff(E) {l-exp [-t A eXP(-EfRT)]} dE
E .
mln
*
According to this equation,W
v
+ W
v
as t + 00, independently of temperature.
However since the pyrolysis time is limited in practice, the calculated ultimate
weight loss will be temperature dependent in agreement with the experimentally
observed behavior.
Expression (6.3) can be viewed as an integral equation relating the unknown
function f(E) to the experimentally measured function Wv(t). Pitt solved this
equation by an ingenious approximate technique using weight loss data from the
pyrolysis of a high volatile bituminous coal at temperatures 300-650
o
C and
times 10s-100 min. In these calculations Awas given the value 10
15
s-1
Jecause of experimental limitations the weight loss at very short and very
long times could not be measured accurately, whence the dotted sections of the
curve represent an extrapolation. The weight loss calculated with the curve
f(E) of Fig. 6.1 was in good agreement with the experimental weight loss.
It is informative to seek possible chemical interpretation of Pitt's model.
As discussed in the previous chapter, tar constitutes 75% or more of weight
loss for high volatile bituminous coals, therefore, the sharp peak in Figure 6. I,
at about 50 to 55 kcal, would correspond to the tar species. This behavior
would be consistent with the hypothesis that the main tar forming reaction
involves the dissociation of the ethylene bridge (reaction 05 in chapter 3)
whose activation energy would be in the range 48-57 kcal depending on the size
and the substituents of the aromatic nucleus. It must be emphasized, however,
that it has not as yet been established that the dissociation of the ethylene
bridge constitutes the main elementary reaction in tar production. Similar
Fig. 6.1. Distribution of activa-
tion energies in Pitts' model
(source: ref. 137).
114
-
o
.,A__
I \
, \
interpretation would suggest that
the part of the curve at low E corre-
sponds to the formation of H
2
0 and
CO
2
while the part at high E corresponds
to the formation of hydrocarbon gases,
CO and H
2
For example, reaction D2
which is the main methane forming step
has activation energy about 70. The
identification of the activation ener-
gies involved in the Pitt model with
actual bond breaking energies provides
the satisfaction that the model reflects
some aspects of physical reality.
A further interesting implication of
to match, in a least squares sense, Eq.
to a single first order reaction
W, " w: {l-e>P [-t A e>p (-'IRTI}}
Pitt's model is manifested by attempting
(6.3) with the expression pertaining
(6.4)
The two expressions would not agree very well over a broad range of temperatures,
but in a limited range one can compute the value r that provides the best fit.
The value of E calculated in this fashion turns out to be significantly lower
than the mean of f(E). This behavior can explain to some degree the low
activation energies determined in reference to the single reaction model.
Going back to Figo 6.1 it is noted that curve f(E) has a single narrow peak,
therefore, it could be well parametrized by two or three parameters, e.g. the
location and width of the peak, without seriously compromising the ability to
describe experimental weight loss datao Anthony et al. (refs. 126,136) utilized
a Gaussian form,
f(E) = ~ exp [ __1_ (E-E )2]
(bT) 2
0
20
2
0
(605)
*
containing the two parameters E and o. They also treated W as an additional
o v
unknown parameter rather than identifying it with the proximate volatile matter.
Thus, their model contained four parameters to be estimated from experimental
*
data: W
v
' A, Eo' 0, only one more than the primitive single first order reaction
model. In the original papers of Anthcny et al. (refs. 126,136), the limits of
integration for Eq. (6.3) were taken Emin=O, Emax=co. In a later paper (ref. 125),
115
they used Emin=O but treated E
max
as an additional adjustable parameter. Table 6.1
lists the parameter values determined for two coals. For the lignite they deter-
mined two alternative sets which provided equally good agreement with the
experimental weight loss datao
TABLE 601
Experi menta1 parameter values for the model of Anthony et aL
*
Coal Reference W
v
A
Eo
E
min
E
max
(%)
(s-l)
(kcal/gmole)
Lignite 136 40.6 L07xl0
10
4807 0
a
9.38
Bituminous 125
40.6
37.2
L67xlO13 56
0
3
1.67xlO13 54.8
o
o
10.91
6L4 17.2
The model of Anthony et al. is sufficiently simple for combustion calculations
provided some additional data are available regarding the heating value of the
volatiles. An application to fluidized combustion was given in reference 138.
In spite its practical success, this model might be criticized for its assumption
of independent first order reactions. In later sections of this chapter, this
assumption will be discussed in Some detail.
6.1.3 Several first order reactions for individual products
In connection with the physical interpretation of the Pitt-Anthony model, it
was pointed out that the range of activation energies could be divided into
segments which can be tentatively associated with various classes of pyrolysis
productso This idea leads quite naturallytto describing individual product forma-
tion by independent first order reactions. Amodel of independent first order
reactions, one for each product, was employed by the Bergbau-Forschung group
(e.g. ref. 74). Suuberg et al. (refs. 64, 71, 72) used a more flexible approach
by assuming that certain products can be formed by the breaking of two or more
types of bonds, requiring a corresponding number of reactions. The number of
reactions required for each product could be judged from the shape of the
experimental yield-temperature curves. If each product precursor is recognized
as an independent chemical species, its weight Wi changes according to
*
Wi(O) = Wi
Each product results from one to three reactions, corresponding to different
116
*
values of i and involving independent sets of Darameters (Wi' Ai' E
i
)
Suuberg et al. (refs, 64, 71, 72) conducted detailed measurements of pyrolysis
products for a lignite and a high volatile bituminous coal using the captive
sample technique (Chapter 3). In most experiments, the temperature-time history
was a pulse consisting' of a heating segment with rate about 1,0000C/s leading
to the desired peak temperature. Immediately after the peak temperature was
attained, the current was interrupted and the ensuing cooling segment with rate
roughly - 2000C/s completed the pulse. Under these conditions, changing the
peak temperature also changed the pulse width, i.e. the effective duration of
the experiment. In some experiments, an isothermal temperature segment of 2-10
seconds followed the peak temperature to ensure complete devolatilization. The
*
experiments employing protracted heating were used to determine Wi' while the
experiments using the simple pulse were used to determine the kinetic parameters
Ai' E
i

Tables 6.2 and 6.3 list the parameter values determined for the lignite and
the bituminous coal, Figures 6.2, 6.3 show measured and calculated yields for
the lignite, corresponding to the parameters of Table 6,2, while Figures 6.4,
6.5 compare results for the bituminous coal, using the parameters of Table 6,3.
The agreement between measurements and model calculations is very good for
methane, ethylene and hydrogen (Figure 6.2) which evolve in a stepwise fashion
with increasing peak temperature. The agreement is not as good for carbon
oxides and water (Figure 6.3) where the experimental points seem to fallon a
curve which'is gradually increasing rather than stepwise. Within the conceptual
framework of the model, the gradual evolution suggests a broad distribution of
activation energies that cannot be accurately represented by one, two, or
even three discrete values.
In the case of the bituminous coal (Figures 6.4, 6.5) the agreement between
model curves and experimental results is good considering the inevitable scatter
in the data, The tar yield in this case was calculated using the "evaporative
diffusion model" combining kinetics and mass transfer (see Chapter 5). All
other products were treated by pure kinetics,
In evaluating the independent reaction model, it should be kept in mind that
the data exhibited in Figs. 6.2-6.5 involve total yields for given T-t pulses.
No time-resolved data were available. The determination of the rate constants
on the basis of cumulative yields becomes increasingly inaccurate with increasing
temperature because of a "compensation effect" between the parameters A. and E..
1 1
In other words, different pairs describe the data equally well, provided a change
in Ai is compensated by a commensurate change in E
i
. This behavior explains the
fact that a common Ai factor for all species other than H
2
and tar described sat-
isfactorily the yields from the bituminous coal. It should also be noted
117
TABLE 6.2
Experimental parameter values for individual product formation in lignite
pyrolysis (ref. 71).
Product
co
Tar
*
(A/s-
1
)
Wi
(% of coal) log E. ( kca1/ gmo1e)
1
5.70 11.33 36,2
2.70 13.71 64.3
L09 6,74
42.0
1,77
12.26 44.4
5,35
12.42 59,5
2,26
9.77 5804
0.34 14.21 51.6
0,92 14.67 69,4
0.15 20.25 74,8
0.41 12.85 6004
0.95 16,23 70,1
2,45 11,88 37,4
2.93 17.30 75.3
16.50 13.90 51,4
0.50 18.20 88,8
a Hydrocarbons other than CH
4
, C
2
H
4
and tar
that several of the entries for log A in Table 6,2 are outside the range 12
to 16 pertaining to unimolecular decompositions. Likewise, the activation
energy of 37,4 for the first tar forming reaction is much lower than the bond
energies estimated in Chapter 3. This is not a serious criticism, however, in
view of the great difficulty of multiple-parameter estimation from expressions
containing sums of exponentials. The case of tar deserves special attention.
The attempt to force the data into a single first order reaction results in a
physically unacceptable set of parameters log A = 2.9, E = 13 kcal/mol, in line
with the observation made in connection with the single reaction model. To
improve on these numbers, the tar yields were fitted (reL 71) by assuming a set
of reactions with continuously distributed activation energies resulting in the
parameter estimates log A = 15.4, Eo = 68.9 kcal/gmole,a = 11.4 kcal/gmole which
118
TJ\I3LE 6.3
Experimental parameter values for individual product formation in the pyrolysis
of a high volatile bituminous coal (refo 71).
Product
co
C
2
H
6
C
3
H
6
+C
3
H
8
Tar
*
Wi (% of coal)
004
0.9
004
2.1
0.7
1.8
0.6
0.5
0.2
0.8
0.4
0.4
0.8
0.5
102
0.4
24.0
5.4
LO
2.9
13
17
Ei(kcal/gmole)
40
65
55
65
55
65
55
65
55
40
44
65
55
65
40
55
65
13
35
90
a other hydrocarbon gases
b light hydrocarbon liquids
are much more in line with the rate parameters estimates of Chapter 30
The models of Pitt, Anthony et al. and Suuberg et al. are very similar in
their underlying assumptions. Both consider product evolution by parallel,
*
i n d e ~ e n d e n t and first order reactions operating on an initial amount Wi of
1.5
.6.6,6-
'"
0
'"
CH
4
0
u
'0
1.0
~

:E
'"
.0;
~
0
~ 0.5
0
C2
H
4
. ~
H
2
>-
0
200 400
1000 1200
Fig. 6.Z Yields of CH
4
, C
Z
H
4
, HZ from the pyrolysis of a
lignite to different peak temperatures at 1 atm He and
1000
0
C/s heating rate (source: refs. 63, 64) .
....
..
Peak temperature (oC)
Fig. 6.3. Yields of HZO, CO
Z
' CO from the pyrolysis of a
lignite to different peak temperatures at 1 atm He and
1000
0
C/s heating rate (source: refs. 63, 64).
119
1Z0
30 r - ~ - - . - - - - " - - - ' - - - - ' - - - ' - - - - ' - - - - ' - ~ - ' - - - - - ' - - - - '

"025
o
u
'0
20
L: 15
'" '0:;
~ 10
Peak temperature

Fig. 6.4. Yields of tar and HZO from the pyrolysis of a

bituminous coal to different peak temperatures at 1 atm He
and 1000
0
C/s heating rate (source: refs. 63, 71).
3.0
2.5
0
0
2.0 u
'0
~ 1.5
L:
0-
'"
1.0
~
"0
'"
0.5
>=
0 0
400 600
Peak temperature (Oc)
Fig. 6.5. Yields of CH
4
, HZ from the pyrolysis of a bituminous
coal to different peak temperatures at 1 atm He and 1000
0
C/s
heating rate (source: refs. 63, 71).
121
volatile species precursors. When the reactions employed by Suuberg are assembled
to describe the evolution of total volatiles, the collection of activation energies
fallon a bell-shaped pattern, similar to the truncated Gaussian curve of Anthony
et al.
The experimental studies of the Bergbau-Forschung group (ref. 74) employed a
linearly increasing temperature-time history with rates varying from 10-
2
DC/min
(using an electric oven) to as high as 10
5
DC/min (using an electrical wire-mesh:
the captive sample technique), The rates of production were measured continuously
by mass spectroscopy providing unusually detailed kinetic information. Unfortu-
nately, the production of tar could not be measured i1 this fashion. As an
example of their results, WE ~ o n s i d e r the production of ethane which they
represented by c single reaction, with log A = 9,33, E = 42,7 kcal/mol, estimated
from the kinetic curve corresponding to a heating rate of 2
D
Cimino The para-
meters for heating rates 10
4
DC/min were 9.95, E = 40,9, Although the model
calculations matched the experimental curves quite well (see Fig, 6.6), the dif-
ference in the rate parameters (the rate constants differ by a factor of twelve
at 600C) requires examination. The discrepancy may be due to the inadequacy
of a single reaction to describe the data or to the fact that the temperature-
time history employed is not well suited for kinetic analysis. However, as can
be seen from Figure 6,6 the ultimate yields represented by the areas below the
curves do not differ by more than 15%, therefore, the data are not inconsistent
with the basic premise of independent reactions,
800 400 500 600 700
Temperature (OC)
m = 2x 10-
2
C/min
E = 42.7 keel Imel
k
o
= 2.9 x 10
10
1 min
m =2 C/min
E =42.7 keel/mol
3
III'
k(\= I. x 10 min 4
m = IxlO C/min
E =40.9 keel Imol
k
o
=5.3xI0
11
/min
0.2
Fig. 6.6. Experimental (----) and calculated
(-----) rates of ethane production at differ-
ent heating rates (source: ref. 74).
We have treated in considerable detail the models of independent first order
reactions because of their attractive features, simplicity and successful
122
description of experimental product yields. One additional, and in our view
important, attribute of these models is the generally reasonable values of the
estimated parameters, With some exceptions, the activation energies are in the
range of bond dissociation energies associated with functional groups encountered
in coal (Chapters 2,3). The idea of product formation occuring by the breaking
of SQ2cific bonds is straightforward and reasonable.
Undoubtedly, the network of pyrolysis reactions is much more complicated than
a set of parallel and independent steps. The models to be discussed in the
next section attempt to describe the coupling and competition between reactions
leading to different ~ r o d u c t s o
6.2 COMPETING REACTIONS
The consideration of competing reaction schemes has been prompted largely
by data showing a negative correlation between the ultimate yields of tar and
gases. Under some experimental conditions, an increase in the yield of gases
is accompanied by a decrease in the yield of tar and vice-versa, The observed
negative correlation could be due to the competition between purely chemical
steps or to secondary reactions in conjunction with internal (intraparticle) or
external (film) mass transfer limitations,
In chapter 4, we discussed pyrolysis product yields obtained by the captive
sample or the entrained flow technique under conditions designed to minimize
secondary reactions on the external surface of the particles or the hot surfaces
of the equipment. Even under these conditions, however, external secondary
reactions may adversely affect the yield of tar at high temperatures and long
pyrolysis timeso For example, the slight decline in the yield of tar above
800
G
C in Figures 4.4-4.7 is most likely due to such external secondary reactions.
Anthony et al. (refs. 136,126) have pointed out that in most of the early experi-
mental studies, low heating rates were invariably associated with densely packed
coal particles, while high heating rates always required dispersed particles.
Evidently, the dependence of tar and gas yields on heating rates observed in
these studies cannot be attributed to the heating rate per se but rather to
the larger or smaller external mass transfer limitations which are incidentally
associated with the heating rate.
Intraparticle mass transfer hinders the removal of tar molecules and thereby
increases the extent of secondary reactions in the interior of the particles.
The extent of such internal secondary reactions increases with pressure and
particle size as illustrated by the data presented in chapter 5. Clearly, to
describe such mass transfer effects, it is necessary to consider secondary
reactions in conjunction with the mass transfer process,
When the pressure is very low and the particles finely ground and well dis-
persed, the secondary reactions in the interior of the particle are largely
123
supressed. Under these conditions any competition between the yields of tar and
gases would be due to purely chemical processes and would be governed by the
only remaining operating variable, the temperature-time history. In the experi-
mental work of the MIT group (refsc 63, 64, 71, 72) which devoted particular
attention to eliminating secondary reactions, the temperature-time history was
varied in three different waysc In the bulk of the experiments, the maximum
temperature was varied at constant heating rate. The results of these experi-
ments, which are generally similar to the results presented in Chapter 4, show
yields of tar and gas to increase monotonically with temperature, and thus do not
indicate the presence of competing reactions. The second type of temperature-
time history employed consisted of fixing the maximum temperature while varying
the heating rate in the range 250-10,0000C/s (refs. 126, 63, 64, 71). For both
the lignite and the high volatile bituminous coal that the experiments employed,
the yields of tar and gases were within experimental error independent of the
heating rateo The third variation of temperature-time history consisted of
heating the coal sample to an intermediate temperature, cooling it and then
subjecting it to a second and higher temperature pulse (refs. 63,64, 71).
Again, the yields depended on the maximum temperature and not on the details
of the heating curve. However, this last type of operation was applied only
to the lifJnite.
The experimental evidence leads to two major conclusions. If mass transfer
limitations can be excluded, the product depend on the peak temperature
and the length of time at that temperature rather on heating rate and other
details of the temperature-time history. Future work may reveal some subtler
dependencies but the data available to date can be well described by independent
reaction models. To describe pressure and particle size effects, it is neces-
sary to consider secondary reactions 'in conjunction with mass transfer processes.
The description of combined mass transfer and kinetics is rather involved as dis-
cussed in the previous chapter.
The rest of this section reviews two coupled reaction models. The first model
(ref. 129) specifically addressed to softening bituminous coals considers a
series-parallel reaction scheme that can be naturally coupled with mass transfer
processeso The second model does not include mass limitations but
ascribes the competitive evolution of tar and gases to purely chemical processes,
namely a set of parallel reactions coupled by common reactants.
6.2.1 Model of Unger and Suuberg (ref. 129)
This model can be represented by the reaction scheme below. First of all, it
is assumed that the fraction of coal consisting of the "inactive" macerals (fusi-
nite, micrinite etc.) does not participate in the pyrolysis. The remaining frac-
tion, consisting of the vitrinite, exinite and resinite macerals reacts in the
1Z4
manner indicated below. Water and carbon dioxide evolve by first order reactions
(constants k
il
, k
iZ
) independent of other subsequent steps. In parallel, coal is
converted to an intermediate material called "metaplast," by a first order reac-
tion with constant k
m
. The term metaplast was introduced in an early model by
Chemin and van Krevelen (ref. 139). The conversion to metaplast proceeds by dis-
sociation of bridges between structural units, therefore, the metaplast molecules,
structurally similar to coal, have a distribution of molecular weights. This
distribution is assumed to be normal and its mean and standard deviation are
treated as adjustable parameters remaining constant during pyrolysis.
Gases I
(C0
2
, H
2
0) H
2
Coal k
il
1
Active km kt
Macera Metaplast ----'-- Ta r
Inactive
Macerals
Gases 2
(CO, CO
2
,CH
4
,etc.)
Once formed, metaplast participates in further reactions. The first is poly-
merization to larger molecules, accompanied by evolution of hydrogen. The effect
of this polymerization on the mean and standard deviation of the molecular weight
distribution is not taken into account. The metaplast also forms a variety of
light products: CO
Z
' CH
4
, C
Z
H
4
, C
Z
H
6
, C
3
H
6
, C
3
H
S
' higher hydrocarbon gases and
hydrocarbon liquids by parallel, independent reactions with constants k
iZ
' These
reactions consume hydrogen to saturate the precursor radicals, hence, their rate
is taken to be proportional to the instantaneous content of aliphatic hydrogen
in the metaplast.
The final step is the conversion of metaplast to tar. Tar consists of vaoors
of metaplast fragments, therefore, this step represents the transfer of tar
molecules from the condensed phase (metaplast) to the gas ohase outside the coal
particle. The rate of this mass transfer step is calculated by solving the dif-
fusion problem in the stagnant region surrounding the particle and turns out to
be proportional to the vapor pressure of metaplast molecules. The mass transfer
problem was discussed in some detail in the previous chapter.
125
Reference 129 does not include mathematical derivations but gives the parameter
values used in the calculations and the comparisons with experimental data. While
this model contains a larger number of adjustable parameters, compared to the
independent reaction models, it provides a more fundamental description of the
pyrolysis chemistry. By virtue of including secondary reactions it allows the
description of pressure and particle size effects. The model is a promising
one although it requires further work relative to the formation and repolymeriza-
tion of metaplast and to possible intraparticle mass transfer limitations.
6.2.2 Model of Solomon (refs. 61, 78)
This model describes purely c1emical processes and utilizes the following
assumptions:
(i) At any instant during pyrolysis, coal is composed of two materials, tar-
forming and non-tar-forming, with weights W
T
, W-W
T
, where initially W=W
o
'
WT=W
oT
' This constitutes the vertical division in Figure 6.7.
(ii) The tar-forming and the non-tar-forming materials are in turn composed of
nonvolatile carbon and certain constituents that are precursors to volatile
products e.g. H
2
0, heavy hydrocarbons, light hydrocarbons, etc. These corre-
spond to the horizontal division in Figure 6.7. The mass fractions of these
constituents are the same in the tar-forming and non-tar-forming materials and
are denoted by Y
i
with initial values Y
oi
'
(iii) Pyrolysis consists of two types of reactions denoted by DT and DG ..
1
Reaction DT generates tar molecules by splitting off vertical slices containing
the instantaneous or current composition of the material. Reactions DG
i
produce
volatile products by splitting off horizontal slices of precursors across both
the tar-forming and the non-tar-forming material. Solomon has aptly made the
analogy with a bowl of soup being consummed by removing spoonfuls (OT) and by
evaporation of ingredients (OG
i
).
(iv) The reactions OG. have the same rates in the tar-forming and non-tar-forming
1
materials, therefore, the mass fractions Vi remain equal in the two materials.
Since Y
i
change with time, the composition of evolving tar molecules also varies
with time although it continues being identical to the composition in the remairfing
coa1.
(v) The reactions OT and DG
i
are all independent and obey first order kinetics.
Because the definitions and notation in refs. 61, 78 are somewhat difficult to
follow, the derivations given below employ a different notation. The instanta-
neous weights of the various constituents in the coal undergoing pyrolysis will
be denoted by Wi' In particular W
nvc
will denote the weight of the horizontal
slice representing the nonvolatile carbono This term is somewhat of a misnomer
126
Non Tar Tar
Forming Fraction Forming Fraction
, -xo X
o
~
Tightly Bound CO
H,O
I
a
light Hydrocarbons
I
H,,,y MPh";C'1
Abstracted Hydrogen by Tar
1
I
v;
I
Non Volatile Carbon
I
i
H, I
I
Loosely
Bound C
co,
Fig. 6.7. Precursors of various pyrolysis products
according to Solomon's model (source: ref. 61).
because although nonvolatile carbon is not affected by any of reactions DG, it
is removed as part of the departing tar molecules. The quantity Y
nvc
(W-W
T
)
which represents the weight of nonvolatile carbon in the non-tar-forming material
is constant,
(6.7)
and proportional to the number of "moles" in that fraction. Likewise, \vc
'W
T
is proportional to the number of "moles" in the tar-forming material.
Assumption (v) implies
which under isothermal conditions is integrated to
(6.8)
The composition variable
127
_ Y
i
_ Yi(W-W
T
)
Yi - Y
nvc
- Ynvc(W-W
T
)
is proportional to the "mole fraction" of constituent i in the non-tar-forming
material and in the coal as a whole, since the composition of both materials is
the same. Assumption (v) then implies that under isothermal conditions
dY
i
crt" = - kiY
i
therefore,
Yi Yoi
- --exp (-k.t)
Y
nvc
- Yo,nvc 1
(6.9)
Relations (6.7)-(6.9) allow the computation of all quantities of interest.
Adding (6.7) and (6.9) yields
Y
nvc
W= Yo,nvc (Wo-W
oT
) + W
oT
Yo,nvc exp (-kTt)
therefore, from (6.9)
Wi = Y
oi
(Wo-W
oT
) exp (-kit) + Y
oi
W
oT
exp [-(ki+kT)t]
(6.10)
(6.11 )
The total weight is obtained by
vI = (vJ -vJ T) I Y . exp (-k.t)
o 0 . 01 1
1
summing over i,
+ iJ TLY . exp
o i 01
(6.12)
where ki=O for i=nvc The composition in the char (remaining coal), Yi=W/W
can be calculated from (6.11) and (6.12).
The instantaneous production of constituent i as a gas (and not with tar) is
given by k.W., hence using (6.11) we find the cumulative production of gas
1 1
i as
o
W.dt'
1
Y . (W -W T) [l-exp (-k.t)J
01 0 0 1
+ Y
oi
W
oT
k ~ l k T {l-exp [(ki+kT)tJ}
(6,13)
The ultimate production of gas i (t ->- 00) is
k.
Y . (W W ) + Y . W _1_
01 0- oT 01 oT ki+k
T
Summing over all ifnvc we find the ultimate weight of gases
produced as
" Y.k.
(l-Yo,nvc)(Wo-W
oT
) + W
oT
f k > ~ T
(6.14 )
(6.15)
(6.17)
(6.16 )
128
The ultimate weight of char is obtained from (6.12) as
w = y (w -w )
00 o,nvc 0 oT
The ultimate weight of tar produced is the difference between (W -W ) and the
o
weightofgases produced. Using (6.15) and (6.16) the weightof tar is obtained as
2
Y .k.
W - W ~
oT oT. ki+k
T
1
According to the above expressions the ultimate weight loss is independent of
temperature but the ratio of gases to tar varies with temperature due to the rela-
tive activation energies of k
i
and k
T
. Since the activation energies of k
i
are
generally larger than that of k
T
, the ultimate yield of gases generally increases
with temperature at the cost of the yield of tar.
The various stoichiometric (X , Y .) and kinetic (k.,k
T
) parameters of the
o 01 1
model were estimated by comparisons with experimental data and the estimated values
are given in ref. 61. Although the procedure used in the estimation was not ex-
plained in detail, it probably involved some trial and error in as much as expres-
sions (6.14) giving the yield of gaseous products include stoichiometric as well
as kinetic parameters. The agreement between model predictions and experimental
data may be judged from Figures 4.4 - 4 ~ 1 1 where the solid lines represent model
calculations. The agreement is good in most cases, however, because of the scat-
ter in the data, the comparisons do not provide a critical test concerning com-
petitive product evolution and other assumptions in this model.
In comparing the models of Solomon (ref. 61) and Suuberg et al. (refs. 64, 71,
72) we note that both employ essentially the same number of adjustable parameters.
Although Solomon's model appears more flexible allowing for the competitive evo-
lution of gases and tar, its ability to represent experimental data does not ap-
pear to be superior to t ~ a t of the independent reaction model. The distinction
between the two classes of reactions DT and DG. is useful from the theoretical
1
standpoint, however, the a priori division between tar-forming and non-tar-form-
ing materials of identical composition is clearly empirical. The estimated model
parameters include very low activation energies for reaction DT and several of
the DG
i
reactions. This may again arise from representing several reactions of
different activation energies by a single reaction.
6.3 DETAILED CHEMICAL MODELS
In principle, a detailed chemical model should describe pyrolysis in terms of
coal's functional groups and their elementary reactions. Chaoters 2 and 3 gave a
survey of structural and kinetic information that might be useful for this our-
pose. However, detailed chemical modeling remains hampered by many serious dif-
ficulties. Available structural information continues being tentative and quali-
tative. For example, the relative concentrations of methylene, ethylene and
129
ether bridges, one of the most important structural have not
yet been quantified. A second difficulty is the large number of individual ele-
mentary reactions making the mathematical analysis very unwieldy. A third yet
difficulty is the lack of reliable physical properties needed for characterizing
mass transfer within the pyrolyzing particles. In view of these difficulties the
efforts to develop a reliable model based on functional groups and chemical reac-
tions have made very limited progress. In the first section of this chapter we
outline a recent kinetic model by Gavalas, Cheong and Jain (refs. 65, 133, 134)
and in the second section we discuss possiole directions for future kinetic modeling
6.3.1 A detailed kinetic model (refs_-,-__65 , 133, 134)
This chemical model rests on the following assumptions:
(i) Coals belonging to a broad range of rank, e.g. subbituminous or bituminous,
can be characterized by a common set of functional groups, with different coals
differing by the concentrations of these functional groups.
(ii) At high temperatures the functional groups react by well known free radi-
cal mechanisms.
Since most functional groups are substituents to aromatic nuclei their
reactivity depends on the type of nucleus and the existence of other neighbor-
ing substituents. For example, the two bond dissociation reactions

OH
2 rgrCHi
involve the same group, the ethylene bridge. The second reaction, however, is
much faster than the first on account of the presence in the ortho position of
the activating hydroxyl group. A complete description of the kinetics must dis-
tinguish between those ethylene bridges that are associated with hydroxyl groups
in the ortho position and those that are not. A description at this level of
detail would require consideration of a very large number of chemical structures
and would be completely out of line with experimental information. To simplify
the situation, the following additional assumption is made:
130
(iii) Any given functional group is exposed to an average environment of other
groups, therefore, its reactivity depends only on the concentration of other
groups.
Assumption (iii) is subject to certain exceptions that can be illustrated by
the following example:
-CH
2
-CH
2
-' 7 -CH2o + '- CH
2
o
-CH-CH
2
-' 7 no dissociation
(a)
(b)
The dissociation of the ethylene group is completely suppressed in the presence
of another group, the a rasical. In this case it is necessary to calculate the
fractiJn of ethylene bridges that do not carry an a radical. Similar restrictions
apply in the presence of double bonds. Such restrictions, or constraints, are
taken into account by calculating the configurations of functional groups that
are susceptible to reactions. The concentration of these reactive configurations
is calculated using another yet simplifying assumption:
(iv) At any instant,the functional groups are located completely randomly, sub-
ject to some obvious constraints (e.g. no two radicals or radical and double bond
can coexist on a single carbon atom). To calculate the concentration of reactive
configurations in terms of the concentrations of functional groups requires the
solution of a combinatorial placement problem which is solved by standard tech-
niques akin to those used in statistical mechanics.
Once the concentration of reactive configurations has been calculated, the
rate of generation of all volatile products other than tar can be calculated. To
calculate the rate of tar production, certain additional constraints must be con-
sidered. Tar molecules contain, by definition, one or more structural units (see
chapter 2), therefore, one or more aromatic nuclei. The vapor pressure and diffu-
sion coefficients of these molecules in the condensed phase decrease rapidly with
increasing molecular weight. Thus tar precursor molecules (metaplast) generated
in the condensed phase have very small probability of evolving to the gas phase
unless they contain only one structural unit.
Having placed a molecular size constraint, we note that the rate of generation
of tar molecules is the product of three factors. The first is the concentration
of precursor units connected by a single bridge to the coal phase. The concentra-
tion of precursor units is calculated by a combinatorial placement technique simi-
lar to the one employed for reactive configurations. The second is the dissocia-
tion rate of the bridges. The third factor is the probability that a unit that has
become free by bridge dissociation will be transferred to the gas phase,rather
than participate in further reactions in the coal phase.
The last factor is quite complicated as it involves mass transfer processes
in the condensed phase and the pore space. Considering first the condensed
phase, a free unit produced by a dissociation reaction such as
131
~ - C H 2 - C H 2 - ~ ' ~ ~ - C H 2 ' + ~ ' - C H 2 '
is subject to immediate recombination by the reverse reaction (cage effect) un-
less it can quickly transfer to the pore space. Likewise,' a free unit formed by
H' + ~ - C H 2 - ~ ' ~ ~ - C H 2 ' + ~ ' H
must diffuse to the surface of a pore or a bubble before escaping to the gas
phase. During this diffusion process, the unit is subject to recombination with
free radical sites on the coal matrix. Because these phenomena are very diffi-
cult to quantify, the model introduces an empirical adjustable parameter Xde-
scribing the probability that a free unit formed by dissociation reactions (meta-
plast molecule) will be transferred to the gas phase rather than recombine in the
coal phase. In some computer calculations Xwas given values in the range 0.1-0.2.
Once in the pore phase, a free unit was assumed to transfer outside the particle
with no further diffusional limitations. This very crude way of handling intrapar-
ticle mass transfer fails to account for the effects of pressure and particle size
discussed in chapter 5.
Having defined the main physicochemical assumptions we can summarize the
model by the block diagram of Fig. 6.8. Blocks 1,2 define the functional groups
whose concentrations characterize the state of the coal at any instant during
pyrolysis. Some sample calculations discussed in ref. 102 employed fifteen
functional groups including two radicals, alpha and beta. Block 3 includes the
calculations of the concentrations of reactive configurations referred to
previously. The information about reaction mechanisms and rate parameters is
introduced through block 4 which in combination with block 3 derives expressions
for the rates of individual reactions (block 5)0 Block 6 includes the calcula-
tions of the rate of tar production. The rate of change of any given functional
group consists of a direct term, due to one or more elementary reactions, and
an indirect term due to the loss of tar molecules carrying the functional group
under question. These two terms are combined in the differential equations
describing the evolution of the state variables (block 7).
Figure 6.9 is a sample of model calculations compared with a few experimental
points for a high volatile bituminous coal. The kinetic parameters used in the
calculations were specified as follows. The initial concentrations of functional
groups were estimated from elemental analysis and nmr data of pyrolysis tars.
Most kinetic parameters were taken from the literature or estimated by thermo-
chemical kinetics while a few parameters were adjusted to get reasonable agreement
with experimental data. The adjusted values were in all cases within physically
reasonable limits.
Using this set of parameters as a base case, sensitivity calculations were
carried out varying one parameter at a time and recomputing the product yields.
132
Elementary Reactions
Rate Parameters
Mechanism of Formation
of Tar and Gases
Fig. 6.8. Organization of the model of Gavalas et al.
(refs. 133, 134).
50 30
+ weigh! loss
40 x to,

CH,
0 ..
C
2
H
s
-0 0
20
u

C
2
H",
0
u
>.
30
"0
;:-
"0
'0
'0
~
0
~
0
"0
1.0
~
OJ
>=
OJ
>=
o ~ ~ ~ ~ ; = : ! : ~ = = = ; i ; ; : = ~ = ~ 0
o w ~ ~ ~ W
Time (5)
Fig. 6.9. Calculated (-----) and experimental product yields
from the pyrolysis of a hvc bituminous coal, "Kentucky No.9,"
at 510
0
C and 1 atm He (source: ref. 134).
133
100 20
A l:l
.2
B
:c
~ . 15 0-
.2
.0;
B
~
A
'0
~
a
10
~
~
&
0
0-
t
~
0
.0
5
<;
20 u
e
u
>,
r
0 0
0 10 20 30 40 50 60
Time (5)
Fia. 6.10. Calculated effect of temperature-time history on
relative product yields from the pyrolysis of an hvc bitumi-
nous coal, "Kentucky No.9, II at 1 atm He. A: isothermal at
600
0
C; B: 400
0
C to 600C linear in 2s followed by isothermal
at 600
0
(source: ref. 65).
The parameter with the greatest effect on the results was the rate constant for
bridge dissociation (reaction 05, Ch3). Other important parameters were the
average number of bridges per structural unit (extent of cross linking) and the
rate constants for radical dissociation reactions (OBI-0B4, chapter 3).
To test the effect of temperature-time history, calculations were performed
for two different conditions. One was isothermal pyrolysis at 600
0
C, The
other consisted of a period of two seconds of linearly rising temperature, from
400 to 600
0
C, followed by isothermal operation at 600C, The isothermal opera-
tion resulted in a slightly higher weight loss (about 2.5%), slightly higher
yield of tar and slightly lower yield of gases, Figure 6.10 compares relative
product yields under the two temperature-time histories. The difference in the
relative yields is slight, especially for tar. Before drawing more general con-
clusions, calculations should be made for several ot:1er types of temperature-
time his tori es ,
In view of the use of several adjustable parameters, the comparison between
calculated results and the few experimental points (Fig, 6.9) have very little
significance, other than indicating the plausibility of the reaction scheme
and pointing out the main pathways for the production of various products. As
formulated in refs. 133, 134 the model suffers from two serious defects. One
134
is the complexity of the rate expressions resulting from the combinatQrial
calculation of reactive configurations. The other is the failure to simplify
the reaction network by eliminating a number of relatively unimportant functional
groups and reactions. The possibilities for such simplifications have emerged
from recent structural and model compound studies but have yet to be exploited
in kinetic modeling. The next subsection will thus be limited to a qualitative
discussion of these possibilities.
6.3.2 Further ideas on kinetic modeling
Recent structural and kinetic studies on whole coal, coal-derived liquids and
coal-like model compounds have provided a sharper focus on the functional groups
and chemical reactions most significant in pyrolysis. This newer information
suggests a revision and simplification of the list of functional groups and reac-
tions employed in refs. 133, 134. By restricting the number of functional groups,
it is no longer essential to utilize random distributions, hence, the rate expres-
sions can be considerably simplified. At the same time the hydroaromatic struc-
tures attain a more crucial status in the overall reaction network. The following
is a revised list of functional groups and chemical reactions that might be suit-
able for future modeling efforts.
functional groups
To attain a manageable model it is essential to keep the number of functional
groups at a minimum. It is thus almost necessary to consider an average aromatic
nucleus, common to all structural units. The average nucleus would contain a cer-
tain number of aromatic carbons and hydrogens and heteroaromatic oxygen sulfur
and nitrogen. These numbers are not integers because they represent not any par-
ticular ring system, but an average ring system.
The substituents on the aromatic nucleus may be limited to methyl and phenolic
hydroxyl. The exclusion of aliphatic chains longer than methyl is suggested by
the work of Deno and coworkers (refs. 12-14) but may be inappropriate for certain
coals.
Hydroaromatic structures are very important and must be carefully considered.
The four groups below have been tentatively suggested in recent structural
studies.
2
3 4
135
These groups have not been directly identified but have been inferred from the
interpretation of various analytical data. The selective oxidation studies of
Deno and coworkers suggest the absence of unsubstituted tetralin structures,
although they allow for the possibility of methyl-substituted tetralin structures
(ref. 12). Likewise, structure 2 was suggested as predominant in an Illinois
No.6 coal, while structure 3 was only indicated as possible. Structure 4 has
been employed in order to explain the relatively low ratio of aliphatic hydrogen
to aliphatic carbon determined from 'H and 13C nmr spectra of coal extracts (ref.
140). Other complex multiring structures can also be employed for this purpose
Some recent unpublished functional group analysis of coal derived liquids by this
author also strongly suggests the presence of complex hydroaromatic structures
like 4.
The final type of groups are the bridges. Most recent publications suggest
methylene (-CH
2
-) and diaryl ether (-0-) as the most abundant bridges between
aromatic nuclei. In this respect it must be noted that structure 2 above can be
considered either as a hydroaromatic structure or as two methylene bridges. How-
ever, in the process of pyrolysis it is unlikely that both bridges could be dis-
sociated before other reactions take place. Along the same line of thought, the
biphenyl bond is not meaningfully considered as a bridge because it does not dis-
sociate by thermal means. Finally, ethylene bridges -CH
2
-CH
2
- have not been spe-
cifically identified or excluded on the basis of analytical data.
chemical reactions
The volume of evidence favors free radical rather than pericyclic (concerted)
mechanisms, although the latter may play some limited role. One reaction that
probably occurs by a non-radical mechanism is the condensation of phenolic groups.
Limiting attention to free radical mechanisms we first consider the set of
propagation reactions which determine the relative yield of various products.
Prominent among these are the reactions of hydrogen atoms and methyl radicals.
The hydrogen atoms participate in two types of reactions, hydrooen abstraction
and addition illustrated by the examples
H" + PhCH
3
7 H
2
+ PhCH
2
" (r1)
H + PhCH
3
+ PhH + CH
3
' (r2)
H + PhCH
2
Ph' ~ PhCH
2
" + Ph'H (r3)
Reactions (r2) and (r3) have already been discussed in section 3.6. They are impor-
tant steps for bridge dissociation (r3) and methane formation (r2). Methyl radicals
react quite similarly, e.g.
CH
3
+ PhCH
3
+ CH
4
+ PhCH
2
"
CH
3
+ PhCH
2
Ph' 7 PhCH
2
" + Ph'CH
3
Alpha radicals participate in hydrogen exchange reactions such as
(r4)
(r5)
136
.
PhCHi +
()QJ

()QJ
+ PhCH
3
!
(r6)
rOO
+ H
Reaction (r6) serves to saturate alpha radicals and at the same time regenerates
hydrogen atoms. While the hydroaromatic groups 2,3 react fairly simply as in
(r6) above, the groups 1,4 react in a more complicated way. The propagation
reactions of group 1 (tetralin-like structure) are summarized below, where
the phenyl ring could actually be a naphthyl or other ring system. Reaction
(r7) leads to methylindan while (r8) leads to the relatively unstable dihydro-
naphthalene which eventually ends up as naphthalene (rlO). The addition
reaction (r
ll
) can be followed by various hydrogen abstractions and dissociations
.
ro
.
6
+ H
( r7)
( rS)
R
+ RH +
00.
(rg)
H +
( rID)
D

137
(rll )
(R
resulting in C
2
-C
4
hydrocarbons. The point to be emphasized here is that light
hydrocarbon gases other than methane need not arise from aliphatic chains but may
largely arise from the decomposition of hydromatic structures as illustrated by
reactions (rll), (rI2).
Among various possible initiation reactions, only three are relatively energe-
tically favorable,
Ph-CH -CH - P h ~ ~ Ph-CH " + Ph'-CH
Z
"
Z Z ~ Z
(rI3)
00
.-
00 ::
C
2
H
4
(rI4)
...

+
.
O

00
0
(rI5)
...
where the phenyl ring could be any aromatic ring system. Which of reactions (rI3-
r15) is more important is not known at this time because of uncertainties about
the concentration of ethylene bridges and the magnitude of the cage effect on the
rate of (rI3). The kinetics of reactions (rI4), (rI5) are also poorly understood
as they involve ring opening with formation of birarlicals. As a result of these
uncertaintaies the rate of initiation reactions cannot at this point be predicted
from first principles, even if complete structural information was available.
Termination reactions are due almost entirely to recombination of the rela-
tively abundant alpha radicals. The rate of this recombination is controlled by
diffusion and might be negligible during the period of rapid product formation.
138
Chapter 7
HYDROPYROLYSIS
Heating coal in hydrogen rather than in an inert gas results in a significantly
different product distribution and merits separate consideration. In particular,
the increased production of single ring aromatics makes hydropyrolysis a poten-
tially attractive route to chemicals from coal. The changes in the network of
thermal reactions engendered by the presence of hydrogen can be roughly classi-
fied as follows:
(i) During the e ~ r l y stages of pyrolysis, characterized by rapid tar release,
hydrogen penetrates the coal particle and reacts with various free radicals in
the gas phase or the condensed phase resulting in increased volatiles production.
(ii) The tar vapors react with hydrogen outside of the particles producing aro-
matic compounds of smaller molecular weight and, eventually, methane. These reac-
tions include the degradation of condensed rings to single rings and the elimina-
tion of phenolic hydroxyl and alkyl substituents.
(iii) After the prolific formation of tar and gases has ceased, hydrogen reacts
with active sites on the residual char to produce methane. Initially rapid, this
reaction slows down considerably as the char is thermally deactivated.
Processes (ii) and (iii) correspond to what is normally called hydropyrolysis
or flash hydrogenation or hydrocarbonization. Sometimes (e.g. ref. 69) a dis-
tinction is made between hydropyrolysis, referring to relatively high temperatures
(600-1000
o
C) and short residence times, and hydrocarbonization referring to lower
temperatures (450-600
o
C) and correspondingly longer residence times. Reactions
(iii), on the other hand, are characterized by the term hydrogasification because
they lead to a single product, methane.
In this chapter we will be concerned with hydropyrolysis, i.e. reaction groups
(i) and (ii) at the exclusion of hydrogasification which is more properly dis-
cussed in the context of coal gasification. In sections 7.1-7.4 we examine four
types of experimental systems for hydropyrolysis, the captive sample system, the
packed bed system, the modified captive sample system, and the entrained flow
system. Section 7.5 contains a review of model compound studies that relate me-
chanistically to coal hydropyrolysis. The final section 7.6 reviews kinetic model-
ing of hydropyrolysis.
7.1 CAPTIVE SAMPLE EXPERIMENTS
The apparatus and procedure used in these experiments are the same as the ones
used for straight pyrolysis (Section 4.1.2). In fact, the measurements reviewed
below were part of the pyrolysis program carried out by the MIT group. The cap-
tive sample technique allows relatively rapid removal of volatiles from the
139
reaction zone so that reactions in group (ii) are largely suppressed and hydro-
pyrolysis is essentially limited to reaction group (i). This constitutes a limi-
tation of the captive sample technique from the standpointcof process-oriented
research, where reactions (ii) are utilized to produce the highly desirable single
ring aromatics.
Comparisons between weight loss- or total volatiles - under conditions of pyrol-
ysis and hydropyrolysis have been made by Anthony et al. (ref. 1Z5) and Suuberg
(ref. 63). Figures 5.8, 5.9 and 5.13 in Chapter 5 taken from the work of Anthony
et al. show the weight loss as a function of temperature, pressure and particle
size respectively. In Fig. 5.8, the weight loss under 1 atm He (or N
Z
)' 69 atm
He and 69 atm HZ is the same until about 600
0
C above which the weight loss under
69 atm HZ exceeds that under 1 atm He which in turn exeeds the weight loss under
69 atm He. In these experiments the sample was rapidly heated to its final tem-
perature at which it was maintained for 5 to ZOs.
Figure 7.1 shows the results that Suuberg obtained for the same bituminous
coal using a temperature time history consisting of a sharp pulse (see Section
5.Z). The weight loss for all three atmospheres is identical, within experimen-
tal error, until a peak temperature of about 750
0
C. Above this temperature, the
o
40
;t. 30
10

o I aIm He
69 atm He
69 atm Hz
Heating role IDOOoe /s

Peak temperature (oC)
Fig. 7.1. Weight loss vs. peak temperature for
pyrolysis and hydropyrolysis of a bituminous coal
II Pittsburgh No. 8 (source: ref. 63).
140
weight loss is essentially the same at 69 atm HZ and 1 atm He and exceeds that
at 69 atm He.
In Figures 5.8 and 7.1, the weight loss curves start diverging at a temperature
which marks the transition from conditions free of mass transfer limitations to
conditions limited by mass transfer. The different transition temperatures,
600
0
C in Fig. 5.3 versus 750
0
C in Fig. 7.1 are evidently due to the different
temperature time histories. Increasing the residence time at the highest tempera-
ture, lowers the temperature of transition to mass transfer limitations.
Another effect of the prolonged residence time employed in the experiments of
Fig. 5.3 is the higher weight loss at 69 atm Hz, compared to that at 1 atm He, a
behavior which is not displayed for the pulse-like temperature histories of Fig. 7.1.
This result can be attributed to the contribution of hydrogasification reactions
(group iii) which is substantial only at the longer residence times. The increased
weight loss at the longer residence times due to hydrogasification reactions is
also evident in Figs. 5.10 and 5.13.
The most detailed measurements comparing.product distributions in pyrolysis
and hydropyrolysis were made by Suuberg et al. (refs. 63, 141). Fiqures 7.Z -
7.4 summarize some of their results.
Figure 7.Z compares the tar yields at 1 atm He, 69 atm He and 69 atm HZ' As
we have already seen in Chapter 5 (Fig. 5.11), the yield at 1 and 69 atm He re-
main the same until 700
0
C beyond which the yield at 69 atm drops considerably below
the atmospheric yield. The yield at 69 atm HZ is subject to competing effects. On
25
o I atm He
69 etm He
A 69 atm Hz
o
o
32
Q.>
-:;, 10
o
I--
5

Peak temperature (oC)
Fig. 7.Z. Tar yield vs. peak temperature for pyrolysis
and hydropyrolysis of a bituminous coal "Pittsburgh No.8"
(source: ref. 63).
141
the one hand, hydrogen stabilizes free radicals susceptible to reattachment in
the condensed phase. On the other hand, hydrogen reacts with tar precursors in
the voids or in the coal melt to produce lower molecular weight products and,
at the same time, the increased pressure suppresses the rate of mass transfer
away from the particle. The scatter in the data of Fig. 7.Z does not allow quan-
titative assessment as to the relative magnitude of these effects.
Figure 7.3 compares the yields of methane in 1 atm He and 69 atm HZ' At all
temperatures the yield in hydrogen considerably exceeds the yield in the low
pressure inert environment. The large differences in the yield are evidently
due to the synergism of the two factors mentioned earlier. High pressure reduces
the rate of mass transfer and thus increases the probability of secondary reac-
tions including reactions of hydrogenolysis of tar vapors. Additional con-
tributors to the increased methane yield are reaction of molecular hydrogen with
active sites in the coal matrix that are not associated with tar precursors. Such
reactions include the elimination of methyl substituents on aromatic rings. In
connection with Figs. 7.1 - 7.3 it must be noted that the effects of hydrogen on
tar and gases are in the opposite direction, whence the more modest effect on
total weight loss.
In addition to the bituminous coal, Suuberg studied a lignite with respect to
product yields under conditions of pyrolysis and hydropyrolysis (refs. 63, 141).
As shown on Fig. 7.4, starting with about 500
0
C, the methane yield under 69 atm
HZ exceeds the yields under 1 and 69 atm He. The latter two yields remain equal
until about 700
0
C which marks the inception of mass transfer limitations.
Increased yields of hydrocarbon gases other than methane and ethylene were
similarly observed in the presence of hydrogen at temperatures as low as 500
0
C.
The low temperature marking the deviation between the gas yields from pyrolysis
and hydropyrolysis signifies as before that hydrogen does not only influence the
course of secondary reactions of tar precursors but participates in direct reac-
tions with the coal matrix.
In contrast to the yields of other hydrocarbon gases, the yields of ethylene
at 1 atm He and 69 atm HZ were equal and, beginning at 700
0
C, surpassed the
yield at 69 atm He.
Although the amount of tar obtained from lignite was low and, hence, subject
to larger measurement error, it could be still observed that the tar yields at
1 atm He and 69 atm HZ were higher than the yield at 69 atm He. The weight loss
at 69 atm HZ slightly exceeded that at 1 atm He. Compared to the bituminous
coal, lignite displays a somewhat different weight loss dependence on total pres-
sure and hydrogen pressure probably due to the difference in the relative tar
yields between the two coals.
Peak temperature (OC)
Fig. 7.3. Methane yield vs peak temperature
for pyrolysis and hydropyrolysis of a bituminous
coal "Pittsburgh No.8" (source: ref. 63).
'if.
6
:o! 4
'"
'"
c
o
.<=
Q; 2
::;:
I otm He
o 690tm He
690tm H
2
8

Peak temperature (Oe)
Fig. 7.4. Methane yield vs. peak temperature for pyrolysis
and hydropyrolysis of a lignite (source: ref. 63).
143
7.2 PACKED BED EXPERIMENTS
In this arrangement, the coal sample is held stationary in a section of a
tubular r e a c t o ~ which we shall call the hydropyrolysis section, where it is sub-
jected to.a temperature program under hydrogen flow. The volatile products car-
ried in the hydrogen stream pass through an additional heated section, which we
shall call the hydrogenolysis section, and after quenching are conducted to pro-
duct collection and sampling equipment. By regulating the hydrogen flow rate
and suitably controlling the temperatures in the hydropyrolysis and hydrogenoly-
sis sections of the tubular reactor it is possible, in principle, to control the
temperature and residence time of the solid and the volatile products independently.
In early experiments by Hiteshue et al. (refs. 142, 143) util izing the packed
bed arrangement, the heating period was relatively long, about three minutes, and
the temperatures of the solid sample and the volatile products could not be con-
trolled independently. In a recent study, Finn et al. (ref. 144) used a two-seg-
ment tube to implement independent temperature control of solids and volatiles
while achieving heating times as short as half a minute.
A schematic of the apparatus used by Finn et al. is shown in Fig. 7.5. Coal
was placed in an 8 cm long bed in the hydropyrolysis section and subjected to a
Muffle Furnace
Hydrogenolysis
Section
Power Supply
Cool Sed
Hydropyrolysis
Section
Fig. 7.5. Schematic of two-segment hydropyrolysis
reactor used by Finn et al. (ref. 144).
144
temperature pulse by direct resistive heating of the tube wall using a low volt-
age transformer. The hydrogenolysis section was maintained at constant tempera-
ture by a muffle furnace. The reactor tube was 8 mm 10 and the temperature was
recorded at the tube wall. One disadvantage of using a massive coal sample was
that the true heating period was probably considerably longer than the half min-
ute reported for the tube wall. Another disadvantage was the extensive secondary
reactions of tar vapors and other volatiles on the coal surface before entering
the second section intended for hydrogenolysis.
Figures 7.6 - 7.9 show some of the results of Finn et al. (ref. 144). The
yield of various single ring aromatics vs. peak temperature is shown in Fig. 7.6.
The temperature pulse, common in both reactor sections, consisted of a rising
segment (heating rate 7
o
K/s) immediately followed by rapid cooling (three seconds).
The products consisted of approximately equal amounts of benzene-toluene-xylene
(BTX) and phenol-cresols-xylenols (PCX). Both classes of products passed through
a maximum at a temperature slightly below 1900oK. The maximum yield of BTX + PCX
was about 5 per cent.
5
If'()"
4
\
0
0
u
()"
-
""Totol
0
3
Aromatics
:>!!
0 ()"
1:

:0.
01
"G)
.!
2
"0
ell
f!
>=
PCX
BTX
0
0/
0/
__-.L_---.JL-_-J
900 1100 1300
Peak temperature (oK)
Fig. 7.6. Yield of single ring aromatic products
vs. peak temperature for hydropyrolysis of a bitumi-
nous coal at 150 bar pressure and lIs hydrogenolysis
residence time (source: ref. 144).
145
o 10oK/s
Ii. 20
o
K/s
o 30
o
K/s
o
o
STX ( b)
(a) pex g3.0
u
8
u
'0
2
.
0
-
o
:>!?
o 2.0
:E
C'
Gl
.! 1.0
'0
Gl
;;:

oL-_.L-_.L-_..I.-_..I.-_...L_-L__
o 2 4 6 8 10 12 14
Hyd rogenolysi s time (s)
Fig. 7.7. Yield of BTX and pex vs. hydrogen-
olysis time at different heating rates for
hydropyrolysis of a bituminous coal at 150 bar
o
pressure and 1000 K peak temperature (source:
ref. 144).
Fig. 7.7 plots the product yields obtained under the same type of temperature
pulse as in Fig. 7.6 but for different heating rates and hydrogenolysis times.
The yield of BTX increases steadily with hydrogenolysis time and is rather insen-
sitive to heating rate. These trends can be explained by the fact that BTX is
an intermediate product resulting from the degradation of tar and in turn being
converted to methane. Since the latter reaction is slower, the maximum of BTX
corresponds to hydrogenolysis times larger than ten seconds and is not shown in
the figure. The effect of heating rate is smaller and largely masked by the
scatter in the data.
In contrast to the yield of BTX, the yield of pex shows some rather striking
trends. At fixed heating rate, the yield passes through a maximum at about three
seconds hydrogenolysis time. The presence of this maximum suggests consecutive
reactions from tars to pex and pex to BTX or directly to methane. Since the
decrease in pex is not accompanied by a commensurate increase in BTX, the direct
conversion of pex to methane seems to be the predominant route. At fixed
146
hydrogenolysis time, pex decreases rather rapidly with increasing heating rate
probably due to the shorter soZids exposure to high temperatures decreasing the
yield of precursor tar vapors. Why this same effect is not shown by the yield
of BTX remains a vexing question.
Adifferent temperature-time program was used in the measurements reported in
Figs. 7.8, 7.9. After rising to its maximum value, the temperature in the
hydropyrolysis section was maintained constant for ten to fifteen minutes while
the temperature in the hydrogenolysis section was kept at some other constant
value throughout the run Figure 7.8 shows various product yields vs. hydrogen-
olysis Since the hydrogen flux was kept constant in these runs,
variation of the hydrogenolysis temperature was accompanied by variation of the
hydrogenolysis time. Nevertheless, the yield curves still reflect the fact that
tar vapors are precursors for benzene and other light aromatics which in turn
are converted to the final product methane. Figure 7.9 plots the yields of
several products vs. hydropyrolysis temperature. The maximum yield of benzene,
about 12 percent, is quite promising from the standpoint of producing chemicals
from coal.
7.3 MODIFIED CAPTIVE SAMPLE EXPERIMENTS
To achieve high heating rates and prevent secondary reactions on the coal par-
ticle surface, Graff et al. (refs. 145, 146) developed an experimental technique
20
o
o
u
-
o
10
x'

x
/
Hydrogenolysis temperoture (oC)
Fig. 7.8. Product yields vs. hydrogenolysis temperature
for a bituminous coal at heating rate 1
0
K/s, peak hydro-
pyrolysis temperature 750
0
Kand hydrogen pressure 100 bar
(source: ref. 144).
147
30
Methane x
/
g 20
/'
u
-
0
~

'0
10
~
Benzene
GO
;:
Ethane
0
800 1000 1100
Peak hydropyrolysis temperature (Oel
Fig. 7.9. Product yields vs. peak hydropyrolysis tempera-
ture for a bituminous coal at heating rate 5
0
K/s, hydro-
genolysis temperature 1123
0
Kand hydrogen pressure 150 bar
(source: ref. 144).
/Power Supply
....-----{"v }-----,
H2 -
Cool
Sample
Thermocoupl
,/
Filter
'.- -
'-v---J
Hydrogenolysis
Section
Fig. 7.10. Modified captive sample reactor for coal
hydropyrolysis (source: ref. 145).
combining the advantages of the captive sample and the packed bed techniques.
The reaction section of their setup is shown in Fig. 7.10. The reactor consists
of a stainless steel tube 5.1 mm ro, 6.3 mm 00 and 30 cm length capable of with-
standing up to 1000
0
C temperature and 100 atm pressure. The finely ground coal
is deposited on a circular region in the middle of the tube. The whole reactor
tube is heated resistively by means of a DC power supply switched on and off by
a control circuit. As with the captive sample equiment described in Chapter 4,
resistive heating is applied at two levels. The first and higher level serves
to heat the tube to the preset temperature at a rate up to 1500
0
C/s. After the
desired temperature is established, a control circuit switches power to the lower
148
level, adequate to maintain the tube at the desired steady temperature for the
duration of the experiment. A spot welded thermocouple serves to indicate the
temperature and activate the switching circuit.
After establishing the hydrogen flow at the desired pressure and flow rate,
the power supply is switched on and the volatiles released from the coal sample
are carried in the hydrogen stream through the downstream section of the tube
which constitutes the section for hydrogenolysis. While this experimental setup
provides identical hydropyrolysis and hydrogenolysis temperatures, the sample
heating technique can be applied in conjunction with a two-segment reactor with
separate control of the hydrogenolysis temperature. The residence time in the
hydrogenolysis section is in all cases controlled by the hydrogen mass flow rate,
with due allowance for the volumetric expansion at the reaction temperature.
The reaction products are collected in evacuated tanks from which samples are
drawn for analysis by gas chromatography. An ingenious technique is used to
prevent undue dilution by hydrogen of the product gas. A thermal conductivity
cell detects the level of products in the product stream and only. when this level
is above a preset value is the product stream directed to the sample tanks. The
residual char is determined by oxidation in place and analysis of the carbon
oxides produced. Heavy liquids not detectable by gas chromatography, are repor-
ted as "carbon defi ci t."
Some of the results reported by Graff et al. (refs. 145, 146) for a high vola-
tile bituminous coal (Illinois No.6) are reproduced in Figs. 7.11, 7.12 with
the yields expressed as carbon in the products as a percentage of carbon in the
coal. Figure 7.11 shows the yields of methane, ethane and propane vs. reaction
temperature for fixed solids and vapors residence time. The monotonically increas-
inQ yield of methane is obviously due to the fact that this gas is the final
hydrogenolysis product of tar vapors and hydrocarbon gases. Ethane, on the other
hand as an intermediate product passes through a maximum. The monotonic decrease
of propane might be due to the decomposition of its precursor propyl radicals to
ethylene and methyl radicals, favored at the higher temperatures.
Figure 7.12 plots the yields of BTX and tar versus reaction temperature. The
tar was determined indirectly as the difference between the original carbon and
the carbon in all measured products, including char. The determination by differ-
ence is obviously subject to considerable error. As in the studies discussed in
Section 7.2, the yield of BTX passes through a maximum of about 12 percent. This
yield is comparable to that shown in Fig. 7.8 corresponding to much lower heating
rates. The temperature of the maximum was in both cases about 800
0
e. The simi-
larity of the results in Figs. 7.8 and 7.12 indicates that,isolated from other
operating variables, heating rate has a relatively minor effect on product yields.
Fig. 7.11 Yield of gaseous hydrocarbons vs. temper-
ature for hydropyrolysis of a bituminous coal "Illi-
nois No.6" at 100 atm HZ and 0.6s hydrogenolysis
time (source: ref. 146).
30 15
.!!!
I/)
I/)
I/)
0
0
.D
.D
c
20 10
c
0
0
..Cl
.D
...
...
0
0
<..>
<..>
~
10 5
~

"0
"0
Q)
Q)
>.
a a
>.
-
0
-
<..>

BTX
0 <..>
:::::J
:::::J
"0
0 Tar (by difference) 0
"'0
0
0
...
...
a.. -10 a..
600 700 800 900 1000
Temperature
(Oe)
Fig. 7.12. Yields of BTX and tar (by differ-
ence) vs. temperature for hydropyrolysis of a
bituminous coal "Illinois No.6" at 100 atm
HZ,650
0
C/s heating rate and 0.6s hydrogenol-
ysis time (source: ref. 146).
150
15
20_
II)
" ~
"iii
II)
0
0
.0
.0
C
C
0
0
.0
10
10 -E
....
0
0
0
0
~
~ 0
0
-
"0
"0
Q)
5
oa;
>. :>.

-
CH
4
-
0 0
~ C C
2
H
6
+ C
3
H
e
~
"0 "0
0
0 BTX
0
....

Tar (by difference)

....
Q.
0
-lOa.
0 I 2 3 4 5 6
Hydrogenolysis time (s)
Fig. 7.13. Product yields vs. hydrogenolysis
time for hydropyrolysis of a bituminous coal
"Illinois No.6" at 700
0
e, 100 atm HZ and
650
0
e/s heating time ( source: ref. 146).
Figure 7.13 shows the dependence of product yields on hydrogenolysis time (vola-
tiles residence time) at 700
0
e. The maximum in the BTX yield is attained at about
three seconds. This optimal time would decrease to a fraction of a second at 800
0
e
as inferred from the previous figure. The shallow maximum in the yield of methane
is somewhat puzzling in view of the slow decline of BTX at times larger than three
seconds.
7.4 ENTRAINED FLOW EXPERIMENTS
While the modified captive sample reactor is well suited to fundamental kinetic
studies, the entrained flow reactor is better suited to process development by
being amenable to scale-up to pilot plant units. Entrained flow experiments are
difficult to interpret kinetically because coal particles react continuously dur-
ing their passage through the reactor, therefore, the volatile products have a dis-
tribution of residence times for hydrogenolysis. On the other hand, the entrained
flow reactor operating at steady state generates large samples suitable for the
accurate measurement of product yields. An entrained flow reactor for hydropyrol-
ys is generally has geometry simi 1ar to an entrai ned fl ow p.\Iro1ys is reactor, bu t
must be capable of operating at high pressures.
An entrained flow system was used by Fallon et al. (ref. 147) to study the
hydropyrolysis of a lignite and a subbituminous coal at 700-900
0
e and 500-1,500
psia of hydrogen. The products up to and including BTX were determined by gas
chromatography while the heavier liquids were collected and analyzed at the end
151
900
u 850

-
e
....
;:,
800
-e
....
e
Q.
E
e
750
I-
700
0 8 10
(s)
Fig. 7.14. Locus of maximum BTX yield for hydropyrolysis
of lignite at Z500 psia HZ (source: ref. 147).
0.25
-
. ~
II)
e
0.20
o BTX
.0
Co9 +
C
0
0.15 .0
liquids
....
e
0
~
0.10

32
0.05
e
>=
Cg +
0
700 725 750 775 800 825 850 875
Temperature (OC)
Fig. 7.15. Maximum yields of BTX and C
g
+ liquids vs. tem-
perature for hydropyrolysis of lignite at ZOOO psia HZ
(source: ref. 147).
of each experiment. Several runs with lignite explored the effect of the three
principal variables on the yield of products, especially BTX. At fixed resi-
dence time and pressure, the yield of BTX passes through a maximum in the range
700-800
0
C. Likewise, at fixed temperature and pressure, the yield of BTX becomes
maximum at some intermediate residence time, past which it declines rapidly to
950 900
1000 psi
1500 psi
2000 psi
2500 psi
750
OL-_---.l.__-L__
700
152
--;;; 0.25,-----------------------,
en
o
.0 0.20
c:
o
.0
...
8 0.15

-
"0 0.10
Q)
">'
X 0.05

m
Fig. 7.16. Maximum BTX yield vs. temperature for hydropyrolysis
of a subbituminous coal at various hydrogen"pressures (source:
ref. 147),
0.8-
CI'
'u
0.6-
__0---
BTX_y---- 0
-
/"
Q/
,/ 0
;/
..
';' 0.4-
A
"'- ........ pex
,2' A-
.. ----
a:: 6--
1
-"T- A 1/\
oL..-_..L.-_....L-_-'-_--1._--'L..-_-'---=..JL>._....
o 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
_ 1.0..-----------------,
of!
Severity function (dimensionless) .
Fig. 7.17. Relative yields of BTX and pex in the
liquid products as a function of severity for the
hydropyrolysis of a lignite at ZOOO psia HZ (source:
ref. 148),
zero. On the other hand, at fixed residence time, the yield vs. temperature curve
is rather broad around the maximum value. The maximum yields under most condi-
tions were in the range 8-10 percent in terms of carbon conversion. Figure 7.14
is a locus of temperature-residence time conditions under which the BTX yield was
near its maximum value of 8-10 percent. The maximum yields are shown in Fig. 7.14
153
to increase as the hydrogen pressure increased from 500 to 2000 psia. Upon fur-
ther increasing the pressure to 2500 psia, the maximum yield remained essentially
unchanged but the residence time required to attain this yield decreased.
Liquids heavier than BTX (C
9
+) were also observed in significant yields as
shown in Fig. 7.15. These liquids were much lighter than pyrolysis tars consist-
ing of about 40 percent naphthalene and only trace amounts of phenols. However,
being considerably more reactive than BTX, they readily declined with increasing
temperature.
Hydropyrolysis experiments performed using a subbituminous coal resulted in
BTX yields as high as 15 percent compared to the 10 ptrcent obtained with lig-
nite. Figure 7.16 shows the ~ ~ x i m i z e d yield o ~ BTX (with respect to r?sidence
time) as a function of temperature at four pressure levels.
In another recent study, Beeson et al. (ref. 148) studied the hydropyrolysis
of a lignite using an entrained flow reactor with controlled axial temperature
profiles. Although the intent was to determine the effect of the heating rate,
the ability to vary the axial temperature profile offers a potentially useful
variable for product optimization.
The results reported are particularly interesting relative to the detailed
breakdown of liquids in the gasoline boiling range into several fractions: BTX,
C
9
+ aromatics, indenes + indans, phenols + cresols and naphthalene. Overall
yields of these liquid products (carbon in the liquids as a fraction of carbon
in the original coal) ranged between 0.07 and 0.15. The mass fraction of phenols
and cresols in the liquids was as high as 0.76 indicating that the phenolic com-
pounds constitute the primary hydropyrolysis products from lignite. The phenolic
products react further to BTX and methane at a rate depending on temperature and
hydrogen pressure.
Figure 7.17 shows the variation of the relative yields of BTX and phenol +
cresol as a function of a severity parameter defined by
f
t
ko
dt
severity
o
where k 9xl0
5
exp (-30,700/RT). The rate constant k was assigned by refer-
o 0
ence to some earlier data on anthracene hydrogasification, therefore, the severity
parameter is a somewhat arbitrary measure of the combined effect of temperature
and residence time.
Two other related hydropyrolysis programs with emphasis on process and hard-
ware development are the Cities Service short residence time hydropyrolysis pro-
gram (ref. 149) and the Rockedyne program (ref. 150). The City Service program
has employed a laboratory scale entrained flow reactor (about 1 Kg coal/hr) while
the Rockedyne program has utilized a process development entrained flow reactor
(about 200 Kg coal/hr). The distinguishing feature of the second reactor is the
154
rapid mlxlng between feed coal and hydrogen, achieved by a "rocket engine" injec-
tor. Results reported to date for a lignite showed maximum BTX yields of about
10 percent (City Service) or total liquid yields of about 30-40 percent (Rockedyne).
7.5 MODEL COMPOUND STUDIES
As mentioned at the beginning of the chapter, hydropyrolysis reactions include
(i) reactions of hydrogen with the condensed phase during the stage of liquids
formation (ii) hydrogenolysis of the vapors in the gas phase to produce PCX
(phenols), BTX and light nydrocarbon gases. The model compound studies discussed
below are useful primarl1y in understanding the mechanism and kinetics of hydro-
pyrolysis reactions in class (ii) which we have earlier labelled as hydrogenoly-
sis. Some general issues that are of particular interest are the mechanisms of
degradation of ring systems, e.g. naphthalene to toluene or benzene to methane;
and the mechanisms of dealkylation and dehydroxylation, e.g. toluene to benzene
or cresol to toluene. In addition to reaction pathways and mechanisms, it would
be valuable to possess a reasonable k i n e t i ~ description of the effect of operat-
ing variables on product yields.
Virk et al. (ref. 151) analyzed existing data on unsubstituted aromatic hydro-
carbons and found that the rates of disappearance of each compound in pyrolysis
and hydrogenolysis were roughly equal, although the products were different. In
the absence of hydrogen, successive condensation and dehydrogenation led to a
final solid product, coke. In the presence of hydrogen, the final product was
methane. Intermediate products with a smaller number of fused rings were not
specifically identified. Based on the approximate equality of the rates of pyroly-
sis and hydrogenolysis they proposed that both reactions have a common rate deter-
mining step, namely the "destabilization" of the aromatic ring. Although the
mechanism of this step was not identified, its rate was assumed to be related to
the ring delocalization energy.
Penninger and Slotboom (ref. 152) reviewed experimental data on the hydrogen-
olysis of several substituted and unsubstituted aromatics. For the case of un-
substituted naphthalene and phenanthrene they concluded that ring cracking occurs
through the formation of an intermediate hydroaromatic compound. For example,
naphthalene is first hydrogenated to tetralin which subsequently decomposes to
various alkylbenzenes. The mechanism of the crucial first step, the hydrogenation
of the unsubstituted aromatic, was not identified. On the other hand, the sub-
sequent hydrogenolysis of the hydroaromatics was explained by free radical
mechanisms.
The hydrogenolysis of hydroaromatics can be illustrated with the reactions of
tetralin which have already been discussed in a different context (Section 6.3.2).
We are here interested in the mechanism of utilization of molecular hydrogen.
One possibility is offered by the reaction
155
R + H
2
1 H + RH
where R is a carbon centered radical, more specifically ~ n alpha radical. Despite
the unfavorable equilibrium ( ~ G is about 18,800 at 300
0
K), this reaction increases
the concentration of hydrogen atoms which can then participate in addition reac-
tions such as
OO+HO
H'.Ho+OO
the presence of molecular
addition and opening of the
The latter reaction may be
The dihydronaphthalene produced can be subsequently hydrogenated to naphthalene
by the same mechanism, via the addition of a hydrogen atom. No mechanism has so
far been proposed for the direct (pericyclic) addition of molecular hydrogen,
although the possibility cannot be excluded.
Increased concentration of hydrogen atoms due to
hydrogen is effective in tetralin decomposition via
saturated ring (Section 6.3.2) and in dealkylation.
illustrated by the example
H- +
with the methyl radical ending up as methane after hydrogen abstraction.
Cypres and Bettens (refs. 47-49) studied the pyrolysis of phenol and cresols
in the absence of hydrogen and proposed non-free-radical mechanisms for these
reactions (see Section 3.7). The mechanisms proposed leave some open questions
and cannot be readily extended to include the effect of molecular hydrogen.
7.6 MODELING
We start by recalling the classification of reactions into groups (i)-(iii)
defined at the beginning of the chapter. Most of the models concerning hydrogen-
coal reactions have been addressed to reaction group (iii) in the context of coal
gasification to methane. Models of this type will not be discussed here since
they are not relevant to the early phases of hydropyrolysis. Reaction groups (i)
and (ii), dominating the early phases of hydropyrolysis, have been considered
in only a few modeling studies, three of which are discussed below.
156
The experimental work of Anthony et al. (refs. 1Z5, 1Z6) and Suuberg et al.
(refs. 63, 141) have demonstrated the effects of inert pressure, hydrogen pressure
and particle size on the total yield of volatiles (see e.g. Figs. 5.4, 5.9, 7.1).
To quantitatively describe such effects which are intimately related to mass
transfer limitations Anthony et al. (ref. 1Z5) proposed the following set of
phenomenological reactions
* *
(7.1) coal + v
1
V
+ vZV
+ S
*
V +
HZ
+ V (7. Z)
*
V + S (7.3)
* *
S
+ HZ
+ V + S (7.4)
*
S + S (7.5 )
*
Coal decomposes to "unreactive" volatiles V, reactive volatiles V and a reactive
*
solid S. The unreactive volatiles consist.of gases such as methane, steam,
carbon oxides, light liquids (e.g. BTX) and heavier products, tar. The reactive
*
volatiles presumably consist of free radicals or other unstable molecules. S
is a reactive solid susceptible to hydrogenation by the fourth reaction while S
is a solid which participates in no further reactions in the time scale of
i nteres t.
*
Mass transfer enters in the problem through a balance for species V in the
voids of the coal particle. Assuming steady state conditions, the mass balance
becomes
(7.6)
*
where c*, c: are the concentrations of V inside and outside the particle, r
1
,
r
Z
' ... are the rates of reactions (7.1), (7.2), .. , and K is a mass transfer co-
efficient. The reaction rates were expressed in first or second order form and
the rate constants were specified numerically to match the experimental data
(ref. 1Z5). From our standpoint it is important to take notice of the fundamental
assumptions or approximations of the model which were (i) the gas space inside
the particle has fixed volume and uniform composition (ii) the concentration of
hydrogen inside the particle is uniform and equal to the outside concentration
*
(iii) all reactive species can be lumped into one, V (iv) the mass transfer co-
efficient is inversely proportional to the pressure but independent of particle
size. The yield of total volatiles calculated on the basis of this model was in
most respects in good agreement with the experimental yield. However, the model
predicted a stronger pressure dependence than experimentally observed while it
failed to predict the observed effect of particle size in the presence of hydro-
gen. Both points of disagreement seem to derive from assumption (ii) and to
157
suggest the existence of a hydrogen pressure drop from the outside to the inside
of the particle.
In a relatively recent study (ref. 135), Russel et al. carried out an elegant
and comprehensive theoretical analysis of hydropyrolysis reactions coupled with
intraparticle mass transfer. They employed a reaction system identical to (7.1)-
*
(7.5) except that reaction (7.2) was assumed instantaneous, therefore V and H
2
disappeared on a reaction front gradually progressing towards the center of the
particle. Mass transfer was described by the "dusty gas" model, taking into
account fluxes due to diffusion and pressure gradients. For this purpose the
coal particle was assumed to possess a stable pore structure, an assumotion which
applies reasonably well to nonsoftening coals but not to sofening coals (see
Chapter 5). The model was nonetheless tested against the pyrolysis and hydro-
pyrolysis data from a high volatile bituminous coal (refs. 125, 126).
Recent hydropyrolysis modeling work by the MIT group was presented by Schaub
et al. (ref. 153). Although this publication gives very few details, the two
basic premises of the analysis can be summarized as follows:
(i) The hydropyrolysis reactions are represented by the scheme below where Mis
the famil i ar by now metap1as t (Chapter 5), Sis a reacti ve soli d termed "semi-
coke" and A is another reactive intermediate in the condensed phase. Step 3 is
the transport of metaplast molecules from the coal melt to the gas phase as tar.
All other steps are chemical in nature.
~
gases
cool
---L..
~
tor
M
~
~
~ 6
/
S + gases
coke ... 9 A
H
2
CH
4
8
(ii) Step 8 requires the diffusion of hydrogen in the condensed phase which is
initially a melt but later becomes a solid, char.
The reaction scheme shown above differs considerably from the one utilized in
refs. 125, 135. It employs a more complex network of consecutive reactions and
differentiates between three products, gases, methane and tar. It also assumes
that hydrogen reacts with a species in the condensed phase rather than the gas
phase and predicts that the tar yield depends on total pressure but not on the
158
nature of the surrounding gas. Unfortunately, the limited experimental data
available (Fig. 7.2) are insufficient to test this crucial prediction. Another
noteworthy feature of the model is the consideration of hydrogen diffusion through
the coal melt, enhanced by the stirring action of the evolving bubbles. Although
some of its detailed assumptions could be ouestioned,this model represents the
most physically realistic effort in hydropyrolysis modeling.
159
REFERENCES
1 J. G. Speight, Appl. Spec. Rev., 5 (1971) 211-264.
2 P. H. Given, Lecture Notes, 1979.
3 H. L. Retcofsky, T. J. Brendel and R. A. Friedel, Nature, 240 (1972) 18-19.
4 D. D. Whitehurst, in J. N. Larsen (ed.), Organic Chemistry of Coal, ACS
5 R. R. Ruberto, D. C. Cronauer, D. M. Jewell and K. S. Seshadri, Fuel, 56
(1977) 17-24.
6 G. L. Tingey and J. R. Morrey, Battelle Energy Program Report, Battelle,
Pac i fi c Northwes t Labora tori es, 1972.
7 G. R. Gavalas and M. Oka, Fuel, 57 (1978) 285-288.
8 K. D. Bartle, T. G. Martin and D. F. Williams, Fuel, 54 (1975) 226-235.
9 K. D. Bartle, W. R. Ladner, T. G. Martin, C. E. Snape and D. F. Williams,
Fuel, 58 (1979) 413-422.
10 S. Yokohama, D. M. Bodily and W. H. Wiser, Fuel, 58 (1979) 162-170.
11 C. E. Snape, W. R. Ladner and K. D. Bartle, Anal. Chem., 51 (1979) 2189-2198.
12 N. C. Deno, B. A. Greigger and S. G. Stroud, Fuel, 57 (1978) 455-459.
13 N. C. Deno, K. W. Curry, B. A. Greigger, A. D. Jones, W. G. Rakitsky, K. A.
Smith, K. Wagner and R. D. Minard, Fuel, 59 (1980) 694-698.
14 N. C. Deno, B. A. Greigger, A. D. Jones, W. G. Rakitsky, K. A. Smith and
R. D. Minard, Fuel, 59 (1980) 699-700.
15 J. D. Brooks, R. A. Durie and S. Sternhell, Aust. J. Appl. Sci., 9 (1958)
63-80.
16 G. Alberts, L. Lenart, H. H. Oelert, Fuel, 53 (1974) 47-53.
17 S. Friedman, M. L. Kaufman, W. A. Steiner and I. Wender, Fuel, 39 t1960) 33-
45.
18 B. S. Ignasiak and M. Gawlak, Fuel, 56 (1977) 216-222.
19 D. C. Cronauer and R. G. Ruberto, Electric Power Research Institute Report
AF-913, 1979.
20 J. K. Brown, W. R. Ladner and N. Sheppard, Fuel, 39 (1959) 79-86.
21 J. K. Brown and W. R. Ladner, Fuel, 40 (1960) 87-96.
22 H. L. Retcofsky, F. K. Schweighardt and 11. Hough, Anal. Chem., 49 (1977)
585-588.
23 K. D. Bartle and J. A. S. Smith, Fuel, 44 (1965) 109-124.
24 A. A. Herod, W. R. Ladner and C. E. Snape, Prepr. Royal Society Meeting on
New Coal Chemistry, Stoke Orchard, 1980.
25 E. M. Dickinson, Fuel, 59 (1980) 290-294.
26 K. D. Bartle and D. LI. Jones, in C. Karr, Jr. (Ed.), Analytical for
Coal and Coal Products, Vol. II, Academic Press, New York, 1978, pp. 103-160.
27 M. Oka, H. C. Chang and G. R. Gavalas, Fuel, 56 (1977) 3-8.
28 Zander and J. W. Stadel hofer, private communication.
29 S. W. Benson and H. E. O'Neal, Kinetic Data on Gas Phase Unimolecular Reac-
tions, NSRDS-NBS 21, 1970.
30 S. W. Benson, Thermochemical Kinetics, Wiley, Ijew York, 1976.
31 S. E. Stein and D. M. Golden, J. Org. Chem., 42 (1977) 839-841.
32 B. D. Barton and S. E. Stein, J. Phys. Chem., (1980) 2141-2145.
33 B. W. Jones and M. B. Neuworth, Ind. Eng. Chem., 44 (1952) 2872-2876.
34 K. J. Laidler, Chemical Kinetics, 11cGraw-Hill, New York, 1965, pp. 199.
35 D. A. Robaugh, R. E. Miller and S. E. Stein, Prepr. Div. Fuel Chem., Am.
36 A. M. North and S. W. Benson, J. Am. Chem. Soc., 84 (1962) 935-940.
37 R. D. Burkhart, J. Pol. Sci. A, 3 (1965) 883-894.
38 J. N. Cardenas and K. F. O'Driscoll, J. Pol. Sci. A, 14 (1976) 883-897.
39 D. F. /1d1illen, W. C. 09ier and D. S. Ross, Prepr. Div. Fuel Chem., Am. Chem.
Soc., 26 No.1 (1981) 181-190.
40 A. A. Zavftsas and A. A. flelikian, J. Am. Chem. Soc., 97 (1975) 2757-2763.
41 J. A. Kerr and M. J. Parsonage, Evaluated Kinetic Data on Gas Phase Addition
Reactions, Butterworths, London, 1972.
42 M. Levy and M. Szwarc, J. Am. Chem. Soc., 77 (1955) 1949-1955.
43 M. Szwarc and J. H. Bink, in Theoretical Organic Chemistry, Symposium,
Butterworths Scientific, London, 1959, pp. 262-290.
160
44 G. H. Williams, Homolytic Aromatic Substitution, Pergamon Press, New York,
1960.
45 L. W. Vernon, Fuel, 59 (1980) 102-106.
46 J. D. Brooks, R. A. Durie and S. Sternhell, Aust. J. Appl. Sci. 9, (1958)
303-320.
47 R. Cypres and B. Bettens, Tetrahedron, 30 (1974) 1253-1260.
48 R. Cypres and B. Bettens, Tetrahedron, 31 (1975) 353-357.
49 R. Cypres and B. Bettens, Tetrahedron, 31 (1975) 359-365.
50 P. S. Virk, D. H. Bass, C. P. Eppig and D. J. Ekpenyong, in Whitehurst (ed)
Coal Liquefaction Fundamentals, ACS Symposium Series No. 139, Am. Chem. Soc.,
New York, 1980, pp. 329-345.
51 W. von E. Doering and J. W. Rosenthal, J. Am. Chem. Soc., 89 (1967) 4534-4535.
52 B. M. Benjamin, V. F. Raanen, P. H. Maupin, L. L. Brown and C. J. Collins,
Fuel, 57 (1978) 269-272.
53 P. Bredael and T. H. Vinh, Fuel, 58 (1979) 211-214.
54 D. C. Cronauer, D. M. Jewell, Y. T. Shah and K. A. Kueser, Ind. Eng. Chem.
Fundamentals, 17 (1978) 291-297.
55 R. T. Eddinger, L. D. Friedman and E. Rau, Fuel, 45 (1966) 245-252.
56 S. Badzioch and P. G. W. Hawksley, Ind. Eng. Chem. Process Desiqn Develop.
9 (1970) 521-530.
57 H. Kobayashi, J. B. Howard and A. F. Sarofim, Sixteenth Symposium (Interna-
tional) on Combustion, The Combustion Institute, Pittsburgh, 1977, pp. 411-
425.
58 S. K. Ubhayakar, D. B. Stickler, C. W. von Rosenberg, Jr. and R. E. Gannon,
Sixteenth Symposium (International) on Combustion, The Combustion Institute,
Pittsburgh, 1977, pp. 427-436.
59 P. R. Solomon, D. G. Hamblen, G. J. Goetz and N. Y. Nsakala, Prepr. Div.
Fuel Chem., Am. Chem. Soc., 26 No.3 (1981) 6-17.
60 D. B. Anthony, J. B. Howard, H. P. Meissner and H. C. Hottel, Rev. Sci.
Instrum., 45 (1974) 992-995.
61 P. R. Solomon and M. B. Colket, Seventeenth Symposium (International) on
Combustion, The Combustion Institute, Pittsburgh, 1979, pp. 131-143.
62 G. R. Gavalas and K. A. Wilks, A.I.Ch.E. J. , 26 (1980) 201-212.
63 E. M. Suuberg, Sc.D. Thesis, Massachusetts Institute of Technology, 1977.
64 E. M. Suuberg, W. A. Peters and J. B. Howard, Ind. Eng. Chem. Process Design
Develop. 17 (1978) 34-46.
65 R. Jain, Ph.D. Thesis, California Institute of Technology, 1979.
66 H. JUntgen and K. H. van Heek, paper presented to the meeting on coal funda-
mentals, Stoke Orchard, 1977.
67 D. W. Blair, J. O. L. Wendt and W. Bartok, Sixteenth Symposium (International)
on Combustion, The Combustion Institute, Pittsburgh, 1977, pp. 475-489.
68 B. Mason Hughes and J. Troost, Prepr. Div. Fuel Chem., Am. Chem. Soc., 26
No.2 (1981) 107-120.
69 J. B. Howard, W. A. Peters and M. A. Serio, EPRI AP-1803, 1981.
70 P. R. Solomon, United Technologies Research Center Report R77-952588-3, 1977.
71 E. M. Suuberg, W. A. Peters and J. B. Howard, Seventeenth Symposium (Inter-
national) on Combustion, The Combustion Institute, Pittsburgh, 1979, pp. 117-
130.
72 E. M. Suuberg, W. A. Peters and J. B. Howard, in Oblad (Ed), Thermal Hydro-
carbon Chemistry, Advances in Chemistry Series, No. 183, Am. Chem. Soc.,
New York, 1979, pp. 239-257.
73 W. Peters and H. Bertling, Fuel, 44 (1965) 317-331.
74 H. JUntgen and K. H. van Heek, Fuel, 47 (1968) 103-117.
75 R. Cypres and C. Soudan-Moinet, Fuel, 59 (1980) 48-54.
76 P. R. Solomon, United Technologies Research Center Report R76-952588-2, 1977.
77 J. H. Pohl and A. F. Sarofim, Sixteenth Symposium (International) on Combus-
tion, The Combustion Institute, Pittsburgh, 1977, pp. 491-501.
78 P. R. Solomon, Prepr. Div. Fuel Chem., Am. Chem. Soc., 24 No.2 (1979) 179-
184.
79 P. R. Solomon and M. B. Colket, Fuel, 57 (1978) 749-755.
161
80 H. C. Howard, in H. H. Lowry (Ed.), Chemistry of Coal Utilization, Suppl.
Volume, IJiley, New York, 1963, pp. 340-394.
81 A. H. Billington, Fuel, 33 (1954) 295-301.
32 A. J. Forney, R. F. Kenny, S. J. Gasior and J. H. Field, I & EC Product
Res o Dev., 3 (1964) 48-53.
83 D. J. McCarthy, Fuel, 60 (1981) 205-209.
84 O. P. Mahajan, M. Komatsu and P. L. Walker, Jr., Fuel, 59 (1980) 3-10.
85 J. F. Jones, M. R. Schmid and R. T. Eddinger, Chern. Eng. Progr., 60 No.6
(1964) 69-73.
86 W. H. Ode, in H. H. Lowry (Ed.), Chemistry of Coal Utilization, Suppl.
Volume, Wiley, New York, 1963, pp. 202-231.
87 O. P. Mahajan, in R. A. Meyers (Ed.) Coal Structure, Academic Press, New
York (in press).
88 H. N. S. Schafer, Fuel, 58 (1979) 673-679.
89 H. N. S. Schafer, Fuel, 59 (1980) 295-301.
90 H. N. S. Schafer, Fuel, 59 (198U) 302-304.
91 R. J. Tyler and H. N. S. Schafer, Fuel, G9 (1980) 487-4S4.
92 O. P. Mahajan and P. L. Walker', Jr., Fuel, 58 (1979) 333-337.
93 Y. D. Yeboah, J. P. Longwell, J. B. Howard and W. A. Peters, Ind. Eng.
Chem. Process Des. Dev., 19 (1980) 646-653.
94 H. D. Franklin, W. A. Peters and J. B. Howard, Prepr. Div., Fuel Chem.,
Am. Chem. Soc., 26 No.2 (1981) 121-132.
95 P. L. Waters, Nature, 178 (1956) 324-326.
96 D. W. van Krevelen, F. J. Huntjens and H. N. Dormans, Fuel, 35 (1956)
462-475.
97 V. T. Ciuryla, R. F. Weimer, D. A. Bivans and S. A. Motika, Fuel, 58 (1979)
748-754.
98 D. Fitzgerald and D. W. van Krevelen, Fuel, 38 (1959) 17-37.
99 A. G. Sharkey, Jr., J. L. Shultz and R. A. Friedel, in R. F. Gould (Ed.),
Coal Science, Advances in Chemistry Series No. 55, ACS, Washington D. C.
1966, pp. 643-649.
100 A. F. Granger and W. R. Ladner, Fuel, 49 (1970) 17-25.
101 R. L. Bond, W. R. Ladner and G. 1. To in R. F. Gould (Ed.),
Coal Science, Advances in Chemistry Series No. 55, ACS, Washington D. C.,
1966, pp. 650-665.
102 D. M. L. Griffiths and H. A. Standing, ibid, pp. 666-676.
103 A. H. Strom and R. T. Eddinger, Chem. Eng. Progress, 67 No.3 (1971) 75-80.
104 Liquefaction and Chemical Refining of Coal, A Battelle Energy Program Report,
Battelle Columbus Laboratories, 1974.
105 Char Oil Energy Development, Office of Coal Research R&D Report No. 73, FMC
Corp., 1974.
106 Occidental Research Corporation, Final Report on DOE Contract No. EX-76-C-
01-2244, 1979.
107 R. Loison, A. Peytavy, A. F. Boyer and R. Grillot, in H. H. Lowry (Ed.),
Chemistry of Coal Utilization, Suppl. Volume, \'Jiley, New York, 1963, pp. 150-
201.
108 L. H. Hamilton, Fuel, 59 (1980) 112-116.
109 L. H. Hamilton, Fuel, 60 (1981) 909-913.
110 To Matsunaga, Y. Ni shiyama, H. S. Wabe and Y. Tamai, Fuel, 57 (1978) 562-564.
111 P. L.Waters, Fuel, 41 (1962) 3-14.
112 N. Y. Kirov and J. N. Stevens, Physical Aspects of Coal Carbonization,
University of New South Wales, Sydney, 1967.
113 P. L. Walker, Jr. and O. P. Mahajan, in C. Karr, Jr. (Ed.), Analytical
Methods for Coal and Coal Products Vol. I, Academic Press, New York,
1978, pp. 125-162.
114 O. P. Mahajan, in R. A. /1eyers (Ed.), Coal Structure, Academic Press,
New York, in press.
115 H. Gan, S. P. Nandi and P. L. Walker, Jr., Fuel, 51 (1972) 272-277.
116 P. L. Walker, Jr., L. G. Austin and S. P. Nandi in P. L. Walker, Jr. (Ed.),
Chemistry and Physics of Carbon Vol. 2, Mat'eel Dekker, New York, 1966,
pp. 257-371.
162
117 E. C. Harris, Jr. and E. E. Petersen, Fuel, 58 (1979) 599-602.
118 A. Cameron and W. O. Stacy, Austr. J. Appl. Sci., 9 (1958) 283-302.
119 N. Y. Nsakala, R. H. Essenhigh and P. L. Walker, Jr., Fuel, 57 (1978) 605-
611.
120 S. P. Nandi, V. Ramadass and P. L. Walker, Jr., Carbon, 2 (1964) 199-209.
121 Y. Toda, Fuel, 52 (1973) 36-40.
122 Y. Toda, Fuel, 52 (1973) 99-104.
123 M. D. Gray, G. M. Kimber and D. E. Granger, Combustion &Flame, 11 (1966)
399-400.
124 D. Anson, F. D. Moles and P. J. Street, Combustion &Flame, 16 (1971)
265-274.
125 D. B. Anthony, J. B. Howard, H. C. Hottel and H. P. Meissner, Fuel, 55
(1976) 121-128.
126 D. B. Anthony, J. B. Howard, H. C. Hottel and H. P. Meissner, Fifteenth
Symposium (International) on Combustion, The Combustion Institute, Pitts-
burgh, 1975, pp. 1303-1317.
127 A. F. Mills, R. K. James and D. Antoniuk, in J. C. Denton and N. Afgan
(Ed.), Future Energy Production Systems, Hemisphere Publishing Co.,
Washington D.C., 1976.
128 R. K. James and A. F. Mills, Letters in Heat and Mass Transfer, 3 (1976)
1-12.
129 P. E. Unger and E. M. Suuberg, Eighteenth Symposium (International) on Com-
bustion, The Combustion Institute, Pittsburgh, 1981, pp. 1203-1211.
130 P. C. Lewellen, M. S. Thesis, Massachusetts Institute of Technology, 1975.
131 D. W. Van Krevelen and J. Schuyer, Coal Science, Elsevier, Amsterdam, 1957.
132 A. Attar, A.I.Ch.E.J., 24 (1978) 106-115.
133 G. R. Gavalas, P. H. Cheong and R. Jain, Ind. Eng. Chem. Fundamentals, 20
(1981) 113-121.
134 G. R. Gavalas, R. Jain and P. H. Cheong, Ind. Eng. Chem. Fundamentals, 20
(1981) 122-132.
135 W. B. Russel, D. A. Saville and M. I. Greene, A.I.Ch.E.J., 25 (1979) 65-80.
136 D. B. Anthony and J. B. Howard, A.I.Ch.E.J., 22 (1976) 625-656.
137 G. J. Pitt, Fuel, 41 (1962) 267-274.
138 G. Borghi, A. F. Sarofim and J. M. Beer, Prepr. A.I.Ch.E. 70th Annual Meet-
ing, New York, 1977.
139 H. A. G. Chermin and D. W. Van Krevelen, Fuel, 36 (1957) 85-104.
140 T. G. Martin and D. F. Williams, Prepr. Royal Society Meeting on New Coal
Chemistry, Stoke Orchard, 1980.
141 E. M. Suuberg, W. A. Peters and J. B. Howard, Fuel, 59 (1980) 405-412.
142 R. W. Hiteshue, R. B. Anderson and M. D. Schlesinger, Ind. Eng. Chem., 49
(1957) 2008-2010.
143 R. W. Hiteshue, R. B. Anderson and S. Friedman, Ind. Eng. Chem., 52 (1960)
577-579.
144 M. J. Finn, G. Fynes, W. R. Ladner and J. O. H. Newman, Prepr. Div. Fuel
Chem., Am. Chem. Soc., 24 No.3 (1979) 99-110.
145 R. A. Graff, S. Dobner and A. M. Squires, Fuel, 55 (1976) 109-112.
146 S. Dobner, R. A. Graff and A. M. Squires, Fuel, 55 (1976) 113-115.
147 P. T. Fallon, B. Bhatt and M. Steinberg, Prepr. Div. Fuel Chem., Am. Chem.
Soc., 24 (No.3 (1979) 52-63.
148 J. L. Beeson, D. A. Duncan and R. D. Oberle, Prepr. Div. Fuel Chem., Am.
Chern. Soc., 24 No.3 (1979) 72-81.
149 M. I. Greene, Prepr. Div. Fuel Chern., Am. Chern. Soc., 22 No.7 (1977) 133-
146.
150 C. L. Oberg, A. Y. Falk, G. A. Hood and J. A. Gray, Prepr. Div. Fuel Chem.,
Am. Chem. Soc., 22 No.2 (1977) 185-196.
151 P. S. Virk, L. E. Chambers and H. N. Woebocke, in Massey (Ed.), Coal Gasi-
fication, Advances in Chemistry Series, No. 131, Am. Chem. Soc., New York,
1974, pp. 237-258.
163
152 J. M. L. Penninger and H. W. Slotboom, in Albright and Crynes (Eds.), Indus-
trial and Laboratory Pyrolysis, ACS Symp. Series, No. 32, Am. Chern. Soc.,
New York, 1977, pp. 444-456.
153 G. Schaub, W. A. Peters and J. B. Howard, Proc. Int. Conf. Coal Sci.,
Dusseldorf, 1981.
164
AUTHOR INDEX
Alberts, G., 6 (16) 159
Anderson, R. B., 143-rf42, 143)
Anson, D., 89 (124) 162
Anthony, D. B., 41,1\2(60) 160; 89 (125,
126), 90, 91 (125), 92 D26), 93
(125),112 (136), 114 (125, 126,
136, 115 (125, 136), 122 (126,
136),123 (126), 139 (125), 156,
157 (125, 126) 162
Antoniuk, D., 97, (127) 162
Attar, A., 104 (132) 162 -
Austin, L. G., 81
Badzioch, S., 39, 41 (56) 160
Bartle, K. D., 5 (8,9, ll-r,-10 (23),11
(8, 26), 12 (23) 159
Bartok, W., 44, 51, (67) 160
Barton, B. D., 23 (32) 159
Bass, D. H., 36 (50) 16-0
Beer, J. M., 115 (138r-r62
Beeson, J. L., 152, 153-rf48) 162
Benjamin, B. M., 37 (52) 160 -
Benson, S. 19 (29, 30-r,-20 (29), 21,
22, 23 (30), 25 (29, 30, 36), 26
(29),28 (29, 30), 31, 33 (30), 37
(29), 38 (30) 159
Bertling, H., 51
Bettens, B., 35, 36, 155(47, 48, 49) 160
Bhatt, B., 150, 151, 152 (147) 162 --
Billington, A. H., 61 (81) 161-
Bink, J. H., 32, 33 (43) 15-9--
Bivans, D. A., 67 (97) 16-1-
Blair, D. W., 44, 51, (67) 160
Bodily, D. M., 5 (10) 159 -
Bond, R. L., 68 (101) I6f
Borghi, G., 115 (138) 162
Boyer, A. F., 77 (107)-r61
Bredael, P., 37 (53) 16-0-
Brendel, T. J., 4 (3)-r59
Brooks, J. D., 6 (15) 159; 34,65 (46)
160 -
Brown-;-J. K., 9, 10 (20, 21) 159
Brown, L. L., 37 (52) 160 -
Burkhart, R. D., 25 (3rr-159
Cameron, A., 81 (118) 162
Cardenas, J. N., 25 (3sr-159
Chambers, L. E., 154 (151r-r62
Chang, H. C., 11 (27) 159 -
Cheong, P. H., 106, 129-r133, 134), 131
(134), 132, 133, 134 (133, 134) 162
Chermin, H. A. G., 124 (139) 162 -
Ciuryla, V. T., 67 (97) 161 -
Colket, M. B., 41, 43, 4;;r:-47 (61), 56,
58 (79), 125, 126, 128 (61) 160
Coll ins, C. J., 37 (52) 160 -
Cronauer, D. C., 4 (5), 8\19) 159; 37
(54) 160 -
Curry, K. 6, 7, 117 (13)
Cypres, R., 35, 36 (47,48,49),51 (75),
155 (47, 48, 49)
Jeno, N. C., 6, 7, 134 (12, 13, 14),
135 (12) 159
Jickinson, E. 11, 12 (25) 159
Jobner, S., 146, (145, 146),
148 (145, 146) 149, 150 (146) 162
Jormans, H. N. M., 66, 67 (96) 161 ---
Duncan, D. A., 152, 153 (148)
Durie, R. A., 6 (15) 159; 34,
160 -
Eddinger, R. T., 39, 41 (55) 160; 63
(85), 70, 71 (103) 161 -
Ekpenyong, D. J., 36 (50r-r60
Eppig, C. P., 36 (50) 160-
Essenhigh, R. H., 82,
Falk, A. Y., 153 (150) 162
Fallon, P. T., 150, 151-;-152 (147) 162
Field, J. H., 61 (82) 161 ---
Finn, f1. J. 143, 144, ill, 146, 147
(144) 162
68 (98) 161
Forney, A. J., 61 (82)
Franklin, H. D., 66 (94;-161
Friedel, R. A., 4 (3) 159-;-68 (99)
161 -
Friedman, L. D., 39, 41 (55) 160
Friedman, S., 8 (17) 159; 143-rf43)
162 --
Fynes-;G., 143, 144, 145, 146, 147
(144)
Gan, H., 80, 81, 82 (115) 161
Gannon, R. E., 39 (58) 160-
Gasior, S. J., 61 (82) I6f
Gavalas, G. R., 4 (7), rr-(27) 159;
41, 43 (62) 160; 56, 59, 6OT7)
159; 88, 91 \62) 160; 91 (7)
159; 91, 92, 94 (62) 160; 106
\TI3, 134) 162; 106, 107, 110
(62) 160; 129(133, 134), 131
(134)-;-132, 133, 134 (133, 134) 162
Gawlak, M., 8 (18) 159
Given, P. H., 3, 9 T2T 159
Golden, D. M., 21 (31) 159
Goetz, G. J., 39, 40, 160
Graff, R. A., 146 (145, 146), ill (145),
148/ (145, 146), 149, 150 (146) 162
Granger,/A. F., 68, 69 (100) 161 -
Granger, D. E., 89 (123) 162 -
Gray, J. A., 153 (150) 16--2-
Gray, M. D., 89 (123)
Greene, M. I., 106 153, 154
(149), 157 (135) 162
Griegger, B. A., 6, 7,lT? (12, 13, 14),
118 (12) 159
Griffiths, D. 69 (102)
Grillot, R., 77 (107) ill
Hamblen, D. G., 39, 40, 41 (59) 160
Hamilton, L. H., 78 (108, 109) 161
Harris, E. C., 81 (117) 162 ---
Hawksley, P. G. W., 39,
Herod, A: A., 10, 11, 12 (24) 159
Hiteshue, R. W., 143 (142, 143;-162
Hood, G. A., 153 (150) 162 ---
Hottel, H. C., 41, 42 (6OT 160; 89 (125,
126), 92 (126), 93 (125), 114 (125,
126), 115 (125), 122 (126), 133
(126), 139 (125), 156, 159 (125,
126), 139 (125), 156, 157 (125,
126) 162
Howard, H.T, 61 (80) 161
Howard, J. B., 39, 40, 41, 42
(60),43, 44 (64),45 (69), 46, 47
(57), Sl (64, 71, 72) 65 (93),
66 (94) 161; 89 (125, 126), 90 (125)
162; 92 \64) 160; 92 (126) 97
\IT) 160; 112\136) 162; 113 (57)
160; TT4 (125, 126, 136), 115 (125,
136) 162; 115, 116 (64, 71, 72), 117,
118 (TIT, 119 (64), 120 (71) 160;
122 (125, 126 162; 123 (63, 71)
160; 123 (126 162; 123, 128 (64,
IT;"" 72), 138 139 (125),
140, 141 (141 , 126, 141),
157 (125, 126, 153) 162
Hough, M., 10 (22) ---
Huntjens, F. J., 66, 67 (96) 161
Ignasiak, B. S., 8 (18)
Jain, R., 43, 51 (65) 106 (133, 134)
162; 129 (65) 160; 129 (133, 134),
ill (134), 132\133, 134) 162; 131,
133 (65) 160; 133, 134 162
James, R. K., 98, 99 (127), 98 (128TI62
D. M., 4 (5) 159, 37 (54) 160 --
Jones, A. D., 6, 7, 14) 159
Jones, B. N., 23, 27, 32 (33)
Jones, D. W., 11 (26) 159
Jones, J. F., 63 (85) T6T
Juntgen, H., 44 (66), 51-;- 115, 121 (74)
160
Kaufman, M. L., 8 (17) 159
Kenny, R. F., 61 (82) 161
Kerr, J. A., 31, 38
Kimber, G. M., 89 (123) 162
Kirov, N. Y., 79, 104 (1T2T ill
Ka bayash i, H., 39, 40, 41, 46, 47, 113
(57) 160
Komatsu, 62 (84) 161
Kueser, K. A., 37 (54) 16C1--
Ladner, W. R., 5 (9, 11), 9, 10 (20, 21),
10, 11, 12 (24) 159; 68 (100, 101),
69 (100) 161; 145, 146,
147 (144 )162
Laidler, K. J.,-z4 (34)
165
Lenart, L., 6 (16) 159
Levy, M., 32 (42) 1sg-
Lewellen, P. C., 104 (130)
Loison, R., 77 (107) 161
Longwell, J. P., 65 (93) ill
Mahajan, O. P., 61, 62 (84),64
(87), 65 (92), 80 (113, 114) ill
Martin, T. G., 5 (8, 9), 11 (8)
11, 12, 135 (140) 162
Mason Hughes, B., 44 (6sr-160
Matsunaga, T., 79 (110) 161
Maupin, P. H., 37 (52) 160
McCarthy, D. J., 61 (83;-T61
rkConnell, G. 1. T., 68 (TOT) 161
Md1illen, D. F., 27 (39) 159
H. P., 41, 42 (6OT 160; 89
(125, 126), 92 (126), 93\125),
114 (125, 126), 115 (125), 122,
123 (126), 139 (125), 156, 157
(125, 126), 139 (125), 156, 157
(125, 126) 162
Melikian, A. 38 (40) 159
Miller, R. E., 24 (35) 159 ---
Mills, A. F., 97, 98, 99\127), 98
(128)
Minard, R. D., 6, 7, 117 (13, 14) 159
Moles, F. D., 89 (124) 162
Morrey, J. R., 4 (6) 15--9--
Motika, S. A., 67 (97;-161
Nandi, S. P., 80 (115), 81 (116),
82 (115) 161; 83 (120) 162
Neuworth, M. B--.,--23, 27, 32 \33) 159
Newman, J. O. H., 143, 144, 145, m,
147 (144) 162
Nishiyama, Y., rg-(111) 161
North, A. M., 25 (36) 15--9--
Nsakala, N. Y., 39, 40--;-41 (59) 160;
82, 83 (119) ---
Oberg, C. L., 153 (150) 162
Oberle, R. D. 152, 153 (148) 162
Ode, W. H., 64 (86) 161
O'Driscoll, K. F., 2s-(38) 159
Oelert, H. H., 6 (16) 159 ---
Ogier, W. C., 27 (39)
Oka, M., 4 (7), 11 (27) 56, 59, 60, 74
(7) 159
O'Neal, H-:E., 19,20,25,26,28,
37 (29)
Parsonage, M. J., 31, 38 (41)
Penninger, J. M. L., 154 (152)
Peters, W., 51 (73) 160
Peters, W. A., 43, 45 (69),
51 (64, 71, 72) 160; 66
161; 92 (64), 97l!i), 115, 116
\64,71, 72), 117, 118 (71),
119 (64), 120 (71), 123, 128 (64,
71, 72), 138 (69) 160; 140, 141,
156 (141), 157
166
Petersen, E. E., 81 (117) 162
Peytani, A., 77 (107) 161 ---
Pitt, G. J., 113, 114 TTI7) 162
Pohl, A. F., 56, 57 (77) 160---
Raanen, V. F., 37 (52) 160
Rakitsky, W. G., 6, 7, ill (13, 14) 159
Ramadass, V., 83 (120) 162
Rau, E., 39, 41 (55) 16--0--
Retcofsky, H. L., 4 (22) 159
Robaugh, O. A., 24 (35) 159 ---
Rosenthal, J. W., 37 (51r-r60
Ross, O. S., 27 (39) 159 ---
Ruberto, R. G., 4 (5)-;8" (19) 159
Russel, W. B., 106, 157 (135) 162
Sarofim, A. F., 39, 40, 41, 46, 47 (57),
56, 57 (77), 113 (57) 160
Saville, O. A., 106, 157 (1m 162
Schafer, H. N. S., 64 (88, 89, 9OT, 65
(91) 161
Schaub, (153) 162
Schlesinger, M. D., 162
Schmid, M. R., 63 (85) 161 ---
Schuyer, J., 104 (131) 162
Schweighardt, F. K., 10-r22) 159
Serio, M. A., 45, 138 (69) 16--0-
Shah, Y. 1., 37 (54) 160 ---
Sharkey, A. G. Jr., 6s-[99) 1G1
Sheppard, N., 9 (20) 159 ---
Sheshadri, K. 5., 4 (sr-159
Shultz, J. L., 68 (99) 161
Slotboom, H. W., 154 (152T 162
Smith, J. A. S., 10, 12 (23r-r59
Smith, K. A., 6,7,117 (13, m159
Snape, C. L, 5 (9, 11), 10, 11, 12(24)
159
Solomon;- P. R., 39, 40, 41 (59),41,43,
44, 47 (61), 47, 48, 49, 50, 52
(70), 53 (70, 76), 54, 55, (70),
56 (70, 78, 79), 58 (79), 92 (70),
125 (61, 78), 126, 128 (61) 160
Soudan-Moinet, C., 51 (75) 160 ---
Speight, J. G., 3 (1) 159 ---
Squires, A. M., 14G (145,"" 146),147 (Wi),
148 (145, 146), 149, 150 (146) 162
Stacey, W.O., 81 (113) 162 ---
Stadel hofer, 13 (28) 159----
Standing, H. A., 69 (TOZ) 161
Stein, S. L, 21 (31),23\32),24 (35)
159
Steinberg, M., 150, 151, 152 (147) 162
Steiner, W. A., 8 (17) 159 ---
Sternhell, S., 6 (15) 159; 34, 65 (46)
160 ---
Stevens; J. N., 79, 104 (112) 161
Stickler, D. B., 39 (58) 160 ---
Street, P. J., 89 (124)
Strom, A. H., 70, 71
Stroud, S. G., 6, 7,117, 1l8(12) 159
Suuberg, E. M., 43, 44 (63, 64), 51
(63, 64, 71, 72), 90, 91 (63),
92 (63, 64), 97 (71) 160; 101,
104 (129) 162; 115, 116(64, 71,
72), 117, TIS (71),119,120 (63,
64), 121 (71), 123 (63, 64, 71, 72)
160; 123, 125 (129) 162; 128 (64,
72), 139, 140, 141; 142 (63)
160; 140, 141, 156 (141) 162; 156
m) 160 ---
Szwarc, M. ,32 (42), 33 (43) 121
Tamai, Y., 79 (110) 161
Tingey, G. L., 4
Toda, Y., 84, 85 (121r:-84, 86, 87 (122)
162
44 (68) 160
Tyler, R. J., 65 (91r-r61
Ubhayakar, S. K., 39 (58) 160
Unger, P. E., 101, 104, (129)
162
Van Heek, K. H., 44 (66), 51, 115, 121
(74) 162
Van Krevelen;-O. W., 66, 67 (96), 68
(98) 104 (131), 124 (139)
Vernon, L. W., 32 (45) 160
Vinh, T. H., 37 (53) 16--0--
Virk, P. S., 36 (50) 160; 154 (151) 1G2
Von E. Doering, W., 3y-r51) 160
Von Rosenberg, Jr., C. W., 3g-r58)
Wabe, H. 5., 79 (111) 161
Wagner, K., 6, 7, 117 TTI) 159
Walker, P. L., 61, 62 (92),80
(113, 115), 81 (116), 82 (115) 161;
82 (119), 83 (119, 120) 162 ---
Walters, P. L., 79 (111) 1G1 ---
Weimer, R. F., 67 (97) 16--1--
Wender, I., 8 (17) 159 ---
Wendt, J. O. L., 44:'51, 56, 57 (67)
160
Whitehurst, D. D., 4 (4) 159
Wil ks, K. A., 41, 43, 88,91, 92, 94,
106, 110 (62) 160
Williams, D. F., 5 11 (8)
11, 12, 135 (140) 162
Williams, G. H., 32 (44)-r60
Wiser, W. H., 5 (10) 159 ---
Woebocke, H. N., 154 \f51)
Yeboah, Y. D., 65 (93) 161
Yokohama,S., 5 (10) 15g--
Zander, M., 13 (28) 159
Zavitsas, A. A., (40) 159
SUBJECT INDEX
Bubbles
--, role in intraparticle mass transfer
103, 104
COED process 70-72
Free radicals
formation 19-27, 130, 131, 137
--, dissociation 27-29
--, recombination 29-31
--, addition to double bonds 31-32
addition to aromatic rings 32-34,
137, 155
--, hydrogen abstraction 30, 31,
135-137
Functional group analysis 9, 12
--, by linear programming 12-18
Functional groups 3, 14, 134
aromatic nuclei 3, 8, 9, 134
hydroaromatic structures 4, 134
bridges 6, 8, 130, 135
phenolic hydroxyls 6, 8, 27, 34
ether bridges 6, 8
carboxylic groups 8, 34
Gas chromatography for pyrolysis products
41, 43, 44
Heat transfer in pyrolysis 77
--, theoretical analysis 93-99
Heating rate
effect on product distribution 122, 123
effect on plastic properties 78
effect on porous structure of char 82
Hydropyrolysis
captive sample 138-142, 146-150
entrained flow 150-154
packed bed 143-146
processes 153
Mass transfer in pyrolysis 77
--, fil m 99-102
--, intraparticle 102-110
Model compound studies 35-37, 154-155
Models of hydropyrolysis 155-158
Models of pyrolysis
independent first order reactions
111-122
--, competing reactions 122-128
chemical models 128-137
Molecular weight of coal molecules 9
167
Nitr0gen in char and tar 56, 57
Nuclear magnetic resonance spectroscopy 3
-, 5, 9, 11, 56, 58, 60
-, C 5, 9, 11, 56, 58, 60
Research Corp. process 72-76
Plastic properties of coal
in pilot plant operation 75, 76
in mass transfer 79, 80, 102-104,
157, 158
Pore size distribution in coal and char
82-88
Porous structure of coal 80
, changes during pyrolysis 80-88
-, and intraparticle mass transfer
105-110
Pretreatment
--, by oxidation 61, 62
--, by cation exchange 64, 65
Product distribution in hydropyrolysis
140-153
-, effect of pressure 140-142, 153
--, effect of particle size 93
Product distribution in pyrolysis 49-53
effect of pressure 91-93, 101
effect of particle size 92, 93, 101
effect of cation exchange 65
effect of coal rank 49, 51
effect of oxidative pretreatment
61-62
effect of mineral additives 66
--, effect of atmosphere 63, 74
Pyrolys i s
captive sample 41, 44
entrained flow 39, 44
fluid bed 70-72
-,laser 68
light 68
plasma arc 69
processes 70-76
Pyroprobe 44
Rate parameter estimation
activation energies 20, 22, 23, 28,
29
--, A-factors 23-27, 28-33
Reaction of pyrolysis
--, bond dissociation to two radicals
19
168
Reaction of pyrolysis (continued)
--, dissociation of free radicals 28
--, recombination of radicals 29
--, hydrogen abstraction 30
--, addition of radicals to double
bonds 31
--, addition of radicals to aromatic
rings 32
--, of carboxyl groups 34
--, of phenolic hydroxyls 34
--, concerted reactions 34-37
Rheological properties of coal 77-80
Secondary reactions 39, 41, 71, 74,
122, 125
Structural analysis 4, 9-11
Sulfur in char and tar 53-55
Thermogravimetric analysis 66-68
Weight loss in hydropyrolysis 139-154
--, effect of pressure 139-142, 152
--, effect of particle size 93, 156
Weight loss in pyrolysis 45-49
, effect of pressure 89
--, effect of particle size 92, 93, 99