Report No. SRIC 68-13
KINETIC STUDIES ON THE PYROLYSIS, DESULFUR-
IZATION, & GASIFICATION OF COALS WITH EMPHA-
SIS ON THE NON - ISOTHERMAL KINETIC METHOD
-------
Report No. SRIC 68-13
KINETIC STUDIES ON THE PYROLYSIS, DESULFUR-
IZATION, & GASIFICATION OF COALS WITH EMPHA-
SIS ON THE NON - ISOTHERMAL KINETIC METHOD
Marvin L. Vestal, Allan G. Day, III, J. S. Snyderman,
Gordon J. Fergus son, F. W. Lampe, R. H. Esaenhigh
and Wm. H. Johnston
With contributions from
Charles E. Waring, A. L. Warhaftig, J. H. Futrell,
Albert C. Nash, George W. Brown, Pamela P. Farkas,
Curtis A. Johnston and Arlene E. Weitzman
Final Report September 1968 Revised April 1969
Phase I Contract No. PH 86-68-65
with the
NATIONAL AIR POLLUTION CONTROL ADMINISTRATION
Paul W. Spaite, Chief, Process Control Engineering Program
E. D. Margolin, Chief, New Process Development Unit
Leon Stankus, Contract Project Officer
Scientific Research Instruments Corporation
Baltimore, Maryland
-------
TABLE OF CONTENTS
1. INTRODUCTION
ABSTRACT
GENERAL
II. THEORETICAL BASIS OF NON-ISOTHERMAL
KINETIC MEASUREMENTS
PYROLYSIS IN AN INERT ATMOSPHERE WITH NO BACK
REACTION
PYROLYSIS IN A REACTIVE ATMOSPHERE WITH NO BACK
REACTION
PYROLYSIS IN AN INERT ATMOSPHERE WITH BACK REACTION
PYROL YSIS IN A REACTIVE ATMOSPHERE WITH BACK
REACTION
NON-ISOTHERMAL KINETICS OF THE BACK REACTION IN COAL
PYROL YSIS
III .
EXPERIMENTAL METHODS
TYPE OF COAL
IV .
RESULTS AND DISCUSSION
GENERA L NA TURE OF COAL PYROLYSIS
KINETICS OF COAL DESULFURIZA TION
FORMS OF SULFUR
V.
CONCLUSIONS
VI.
REFERENCES
VI I.
A P PEN D I X (Apparatus and Procedure)
Page
1
i
1
6
7
19
21
26
27
32
35
37
37
58
77
80
83
-------
Figure
la.
lb.
2
3
LIST OF FIGURES
Evolution of CH3SH in a non-isothermal pyrolysis in hydrogen.
Graph for determining the activation energy for first order
reactions from the exper imental parameters.
Schematic di3.gram of experimental apparatus.
Sulfur renlOval factors for coal PYl."'olysis in hydrogen at one
atm os]:.here .
4
Cal:> evolution curv~ fo"!: 50/-:. SL1:'£Ul: ('oa: pyroly;:.ed in hydrogen
showing sul.ft.,i cc,n:.cLIllng specJe.s aDd rnajor hydrocarbons,
hull. f'IJ. se~ Table IY £()1~ f'x:)enD','=-ni:21 det~lils.
~
Ga.s ev.)lution curve £01' 5% sulfur cO"ll pyrolyzed in helium
showing sulfur containing species and ma jor h.ydrocarbons,
Fun N4, see Table V [or e::pe r . ,rJ/::ntal de.taiJ s,
6
I~:;(jmparison of HzS evolution In Hz at.mosphere, Run N3,
w [th HzS e'fOhltlon in He atn'losphere. R '.In N 4.
,.
Hydrogen sulfide e\TolutLe'a IroD') dlff,:;rent particle size
cuts of coke bv 11.011.- i::-othermaJ .....~act'1"')n with Hz at one
atrnosphere and 100 ull/rnin £10'." rate, heating rate 5°CI
In in. Coke was pre pared by p y rol:':; is ':if 5% S coal in He
at 900 U C for one hOLl.I'.
8
Analysis of the hydrogf:ll ::::uJfide exp<:Yl1ilental results on the
basis of separate first-OLder i.r rever;:Hble reactions.
9
HzS evolution curve for diH<:Oi"ent hydrogen flow r3.tes.
parameter identifying the curve is h.vdX'0gen flo"-v rate in
mllrl1in.
The
Page
14
16
34
41
44
45
48
57
60
63
-------
Figure
10
11
12
13
14
15
16
LIS T OF FIGURES
Location of the high temperature peak in the HzS evolution in
hydrogen as a function of mean res idence time, "Y.
Compar ison of the HzS evolution curve for a fast back reaction
with the calculated curve for the case of no back reaction.
Relative peak heights for hydrogen sulfide and helium sampled
from the effluent of a bed of coke for Run N22. The carrier
gas contained O. 103% hydrogen sulfide in helium.
Arrhenius type plot of the data from the direct measurement of
the hydrogen sulfide back reaction with coke from Run N22.
Hydrogen sulfide evolution for non- isothermal pyrolysis of
250 mg pyrite in Hz at one atmosphere, flow rate of 100 ml/min,
and heating rate 50 C 'min.
HzS evolution curve calculated by the indirect method.
Comparison of the theoretical curves for HzS evolution in
hydrogen with no back reaction with the experimental result
from Run N23.
17
Rates for the desulfurization reactions and the HzS back reaction
as a function of temperature in hydrogen at one atmosphere.
18
Forms of sulfur in coke as a function of carbonization tempera-
ture, from Powell Reference 2a.
Al
Block diagram of experimental apparatus.
A2
Detailed des cription of coal pyrolys is gas handling system.
A3
Detail schematic of GLC sampling and gas handling system.
Detail schematic of tandem GLC Plas""""a S t h
- .LU pec rograp .
A4
A5
Schema tic diagram of the ion optics for the s canning mas s
spectrometer.
A6
Diagram of reactor vessel used in high pressure experiments.
Page
64
65
67
69
71a
73
75
76
78
5
6
7
8
9
11
-------
Figure
A7
A8
A9
AIO
LIST OF FIGURES
Schematic diagram of the linear temperature
programmer.
Chromatograph calibration chart for hydrogen sulfide.
(Standard 3.5 ml loop; at Xl) .
Plasmagram of 0.1 psia methyl mercaptan in 3.5 cm3
sample loop.
Membrane separator for mass spectrometer inlet system.
Page
13
18
20
22
-------
Table
I
II
III
IV
V
VI
VII
VIII
IX
X
XI
AI
All
LIST OF TABLES
Page
Summary of Analytical Data on 50"!o and 1% Sulfur Coal.
36
Summary of Data for Isothermal Hydrogen Runs.
38
Summary of Data for Isothermal Helium Runs.
40
Summary of Coal Pyrolysis Experimental Data for
Non-Isothermal Hydrogen Runs at One Atmosphere.
51
Summary of Coal Pyrolysis Experimental Data for Non-
I sothermal Helium Runs.
53
Summary of Coal Pyrolysis Experimental Data for Non-
Isothermal Hydrogen Runs at Different Pressures.
54
Summary of Coke Pyrolys is Experimental Data for Non-
Isothermal Hydrogen Runs on Cokes of Different Mesh
Size.
55
Apparent Kinetic Parameters for Hydrogen Sulfide Evolu-
tion in the Non-Isothermal Pyrolysis of 5% Sulfur Coal in
Hydrogen Atmosphere, Run N1.
61
Back Reaction of Hydrogen Sulfide with Coke.
68
Pyr ite Pyrolys is Experimental Data for Non-Isothermal
Hydrogen Run.
68
Kinetic Parameters for the Desulfurization Reactions.
72
Retention Times of Various Compounds in GLC for Triton
305 Column Programmed from 50 700 C and Helium Flow
Rate, 50m1!.min.
17
Mass Spectrometer Calibrations for the Gases Measured
in this Work.
24
-------
1.
INTRODUC TION
ABSTRACT
The theory of the non-isothermal technique developed by Juntgen
and co-workers (loc cit) for the study of the kinetics of complex
heterogenous reactions has been extended to include reactive flush
gases and back reactions of the products, and has been applied to
experimental studies of pyrolysis, desulfurization, and gasification of
coals in a series of twenty three non-isothermal experiments.
In
addition, facilities were constructed for isothermal experiments and
a series of nineteen runs were conducted, utilizing both fast and slow
heating rates to reach the isothermal operating temperature.
The
variables studied included coal particle size, flush-gas composition,
gas flow rates, and the forms of sulfur in the coal.
The experiments
were performed in specially constructed facilities containing some
novel features required for the successful execution of the non-isothermal
technique.
These included a specially constructed mass spectrometer
and a plasma emission spectrometer, as well as convidntional methods
such as chromatography.
The objectives of the work are to determine
kinetics of desulfurization of coal as an aid to the design,operation, and
the evaluation of new process systems for practical desulfurization .
The data and theoretical analyses developed establish the non-iso-
thermal technique as an important method for progress toward a practical
understanding of the desulfur ization kinetics.
A ser'ies of reaction
types have been identified, and their kinetic parameters measured, during
pyrolysis of coal in a hydrogen atmosphere. These are: (1) reaction of
-i-
-------
volatile organic sulfur with hydrogen; (2) reaction of pyrite with
hydrogen to form sulfide; and (3) reaction of hydrogen with the
sulfide and organic sulfur associated with the fixed carbon.
Sulfate in the coal was reduced to sulfide prior to its transforma-
tion to hydrogen sulfide.
The kinetic parameters were measured
for the back reaction of hydrogen sulfide with coke, a reaction which
can be very significant under certain conditions.
Further work is in progress to apply this powerful method of
non- isothermal kinetics to the study of the desulfurization of a series
of ten bituminous coals.
In subsequent reports the desulfurization
kinetics of a variety of coals will be given and the implications of
these kinetic measurements to the general mechanistic picture of
coal desulfurization will be discussed.
-ii-
-------
GENERAL
The combustion of coal containing sulfur is a well-known source
of air pollution.
The present report is Phase I of a project concerned
with the development of processes for desulfurization of coal during
pyrolys is and gasification.
Some potential advantages of removing
the sulfur prior to combustion may be listed.
In the first place, the
volume of gas to be proces sed is ten per cent or les s of that of the final
flue gases and correspondingly the concentration of sulfur is high, in the
two to five per cent range.
Secondly, the sulfur is present largely as
hydrogen sulfide for which the removal-chemistry is probably more
favorable.
Although utilities are naturally hesitant to enter the chemicals
industry, such a development may become desirable; besides the possi-
bility of obtaining organic chemicals from coal, treatment prior to com-
bustion favors the recovery of sulfur as elemental sulfur which is the
most readily marketable form.
Finally, systems based upon removal of
sulfur prior to combustion are probably more easily integrated with thermal
cycles based upon gas turbines, pres surized boilers, and top heat cycles
which can be more efficient.
A large number of coal gasification methods have been brought to
various stages of development under the justification of various technolo-
gical and economic reasons.
Control of atmospheric pollution, however,
was not a prime consideration in such developments.
A reassessment of
coal gasification technology in the light of pass ible contribution to de-
sulfurization is desirable.
Furthermore, recent advances in the kinetic
theory of heterogenous reactions make possible new laboratory methods
with greater applicability to this problem.
-1-
-------
Phase I of this project emphasizes the study of the kinetics of
1 b t ale with emphasis on
coal gasification techniques on a a ora ory sc
t. ns and on the measurements
the identification of sulfur removal reac 10
of the kinetics and the chemical balances of these reactions, both of
sulfur. Furthermore, the relevant back-
pyritic sulfur and of organic
. the kinetic parameters of
reactions are also studied. By measunng
. ."" 1 d" g back reactions the
all lmportant compehng reachons, lnc u in '
information thus obtained is more useful for scale-up conclusions than
laboratory measurements without the basic chemical kinetic parameters.
Initially the results reported in the literature were checked by a series
of essentially isothermal measurements, then a more sophisticated non-
isothermal technique was used to obtain the kinetic measurements required
in this program.
The immense literature pertaining to the reactions and kinetics of
coal gasification include several recent reviews on the chemistry of coall .
Although these reviews cite some recent significant results such as the
27,28
work of Gorin and coworkers of the Consolidation Coal Company,
much of the work directed towards the practical kinetics of desulfuriza-
2-14
tion is on the order of thirty-five years old. This early work on the
practical kinetics of coal desulfurization is summed up in three papers by
2a, b 3
Powell and by Snow
During the first four months of this project a
laboratory for measurements on the heterogenous reaction kinetics of coal
gasification and desulfurization was designed and constructed. The first
series of experiments was carried out essentially isothermally using both
fast and slow heating rates and reactive and non-reactive gaseous environ-
ments. A coal was studied with very similar composition to that used
many years ago in the studies by Snow3 .
-2-
-------
Results of these experiments using modern techniques were in
good agreement with those reported by Snow in 1932.
This verification
of Snow's results incline us to accept his data from a large range of
temperatures and for several different reactive environments.
During the course of these experiments, it became clear that rapid
progress toward the measurement of the kinetic data needed in this pro-
gram would require the combined use of more sophisticated experimental
and theoretical techniques. The recent new method established in Germany
15-22
by Juntgen and co-workers of non-isothermal kinetic measurements
was clearly the most promising new technique to apply to the measurements
of desulfurization kinetics.
Basically this method consists of studying the
kinetics of coal pyrolysis with continually rising temperatures.
Previously reaction kinetics for coal pyrolysis were generally studied
in a series of constant temperature experiments.
Although tedious, the
constant temperature method may give good results.
Difficulties, how-
ever, are frequently enco.untered with irrevers ible reactions and with com-
p1ex heterogenous systems.
For example, the isothermal measurements
checking Snow's data show a major dependence on heating rate.
Chemical
decomposition occurs erratically, making characterization of individual
reactions extremely difficult.
The non-isothermal method is a better-
controlled and reproducible technique for carrying out such measurements.
More importantly, it provides in a single experiment, data which are
interpretable in terms of detailed kinetic parameters for the major
reactions taking place.
The theoretical treatment, initially developed by Juntgen and coworkers
and extended by us, provides a framework for dis cus sing chemical reactions
under non-isothermal conditions.
It, therefore, permits us to examine
more closely the influence of the simultaneous variation of temperature and
- 3-
-------
concentration on the course and velocity of the reactions. The theory
is used to calculate the dependence of sulfur releasing reactions on
. . t' g frequency factor and
temperature, reactIon order, actIva Ion ener y,
rate of heating. The data are obtained by measurements under condi-
tions of controlled, continuous heating of the system.
The most important result of our studies to date has been the imple-
mentation of this approach in characterizing the complex chemical pro-
ces ses occurring in the desulfurization of coal. In the experimental
program we have constructed an apparatus that allows quasi-continuous
mass spectral monitoring of the various pyrolysis products that arise
during the time in which a solid coal sample (in a stream of flush gas)
is being heated at a known and controlled linear rate. We have extended
the theoretical treatment to include the effects of a reactive flush gas
and back reactions of the gaseous products.
Using this apparatus and
the experimental and theoretical procedures developed and described in
this report, we have established the following general conclusions.
1. As suggested by earlier work2, the back reaction of hydrogen sulfide
( the major desulfurization product) with coke plays an extremely signifi-
cant role in the determination of the level of desulfurization achieved.
We have determined kinetic parameters for this back reaction as well
as for the forward reactions of desulfurization.
Under conditions for
which the forward desulfurization reactions achieve reasonable rates,
the back reactions are extremely rapid.
2. Three separate das ses of reaction have been identified which produce
hydrogen sulfide during pyrolysis of coal in a hydrogen atmosphere.
These reactions have been tentatively identified as: (1) reaction with
the volatile organic sulfur; (2) reaction with the pyrite to form sulfide;
-4-
-------
and (3) reaction with the sulfide and the organic sulfur associated with
the fixed carbon.
Kinetic parameters for each reaction have been
determined.
The sulfate in the coal is apparently reduced to a sulfide
prior to its transformation to hydrogen sulfide.
3.
The achievement of a practical process for the desulfurization of coal
by reaction with hydrogen requires that the back reaction of hydrogen
sulfide with coke be inhibited.
This has been accomplished in the lab-
oratory by using high flow rates of hydrogen over small samples of coal.
Obviously this is not a practical industrial process, and further research
will be required to develop an economically feasible method.
Before proceeding to the rational des ign of a practical proces s for
desulfurization of coal along the lines suggested by the present results
several additional laboratory studies will be required.
First it is nec-
cessary to establish the generality of the present kinetic data before
concluding that these parameters are useful for predicting the behavior
of a wide variety of coals.
Secondly, the use of inexpensive sulfur
adsorbents as a practical method for suppressing back reactions of HzS
should be explored.
Kinetic and stoichiometric measurements on po-
tential absorbents, particularly calcined limestones and dolomites, will
be required.
The use of calcined limestones and dolomites as sulfur
absorbents has been proposed in connection with the Consol COz Acceptor
b . 27,28
Process y Gorm et al . The equilibrium constants for the reactions
pertinent to the use of these materials in desulfurization process has
been calculated by Squires 29; however, little kinetic data is presently
available.
These additional kinetic studies will be accomplished in
Phase II of this project.
- 5-
-------
II.
THEORETICA L BASIS OF NON-ISOTHERMAL
KINETIC MEASUREMENTS
The classical, time-honored method used in a study of the
reaction kinetics of a chemical system consists basically of the
measurement of the extent of reaction as a function of time for a
series of experiments, each at a different but constant tempera-
ture. Unfortunately, this method cannot be applied readily to the
pyrolysis of a complex solid substance, such as a coal sample,
because of the uncontrolled occurrence of a number of chemical
reactions during the time that the sample is being heated to the
desired reaction temperature. A method has been developed
recently by Juntgen and co-workers 15-22 which circumvents this
. .
difficulty by studying the reaction velocity at a constant and known
heating- rate of the solid sample. From theoretical kinetic con-
siderations these workers 15-22 have shown how the usual kinetic
parameters, activation energies and pre-exponential factors, may be
.. ,.
derived from such experiments. Moreover, Van Heek, Juntgen,
and Peters 16 have applied the method to the decomposition of basic
magnesium carbonate and have shown that kinetic parameters are
obtained that are in good agreement with those obtained earlier by
classical, isothermal methods23, 24. In addition, the observed
dependence of the instantaneous reaction velocity on sample heating-
t t" 21
rate was in agreement with the theoretical expec a Ion.
-6-
-------
In the present work we have ,adapted the non-isothermal tech-
nique of Juntgen and his colleagues to our studies of the pyrolysis
of coal,. both in inert and reacting atmospheref!1, with particular
emphasis on the kinetics of formation of gaseous sulfur compounds.
In the discussion of the theoretical basis which follows, we recognize
four distinct pyrolytic situations, namely:
1.
Pyrolysis in an inert atmosphere with no back reaction.
2.
Pyrolysis in a reactive atmosphere with no back reaction.
3.
Pyrolysis, in an inert atmosphere with back reaction.
4.
Pyrolysis in a reactive atmosphere with back reaction.
In addition to these cases that are encountered in coal pyrolysis
experiments, we treat by similar methods the kinetics of the back
reaction when it occurs under conditions that the forward reaction
is negligible.
1.
PYROLYSIS IN AN INERT ATMOSPHERE WITH NO BACK REACTION
In the pyrolysis of coal in an inert atmosphere we are concerned
with the thermal decomposition of a solid to form a gaseous product.
This overall reaction can be represented as
-7-
-------
ASoI.CI
... \3 $6' ....
+ C a'"
( 1)
If the gaseous products are continuously removed with
sufficient rapidity the back reaction may be neglected. Such
a decomposition reaction (1) is often first order with respect
to the s olid- pha s e concentration of A, and if this is s Q',
equation may be written as
the rate
dlA1
- (fi=
klA1
( 2)
where [A] is the instantaneous concentration of A in the solid
phase, k is the first order rate constant and t is time.
Continuous
measurement of the solid phase concentration [A) is generally
I I
difficult; however, [A] may be expressed in terms of the volume of
evolved gas C.
That is
I
[Al =
ol (~- V)
( 3)
where V is the volume of gas C evolved up to time t, V 0 is the total
volume of gas C evolved as reaction ( 1 ) goes to completion, and
c( is the proportionality factor. In terms of the volume of evolved
gas the rate equation for the decomposition reaction become s
dV -
dt-
h.(vo -V)
(4 )
-8-
-------
where ~: is the instantaneous rate of production of product gas C
in terms of its volume per unit time.
The first order rate constant k depends upon the temperature
in a manner that is most simply described by the Arrhenius equation
~
~ = k, e- RT
( 5)
where ko is a temperature-independent frequency factor, E is the
activation energy, R is the universal gas constant and T is the
absolute temperature.
For an experimental situation such as con-
sidered here, in which the temperature is not constant but can be made
to vary linearly with the time of reaction, we have
T
-
-
T +
'-
Mt
( 6a)
and
JT -
rt-
M
( 6b)
where M is the heating rate in units of degrees per unit time and Ti
is the initial temperature of the experiment.
The rate equation, (2), may be written in its integrated form
as
-t
tn [A] - \ ~ tit ( 7)
-
-
[A10
0
-9-
-------
where [A]o is the concentration of [A] at zero reaction time.
Under conditions of constant temperature. ( M= 0), k is a constant
independent of t and the right side of (7) integrates simply to kt.
However, for non-zero values of M, k depends on the time and
the integration of (7) is more involved.
Since we know the
relationship between temperature and time to be given by ( 6) ,
we may express the integral in (7) either as a function of time
or of temperature.
It is more convenient to use the latter re-
presentation and so by substitution of (5) and (6) into (7) we
obtain ( 8), viz:
~
LA]
[A1o
-
-
T e
- \to ( e-R"TdT
M )~.
&.
( 8)
where Ti is the minimum temperature at which reaction is ob-
served; that is, Ti is the temperature at which the concentration of
A is [Alo'
forT ~Ti.
Ti is chosen so that, by definition, no reaction occurs
Therefore, we may write equation (8) as
$.,. ~1 - - ~r Ce-I"cIT
[Al 0 - M L 0
+ ~ Te-~drl
T>
L
( 8a)
Since the first integral in the brackets is zero by definition ( T< Ti)
we may write (8a) for mathematical convenience as
~
fA]
lA1C)
-
-
T -
-~(e-~dT
M )0
( 8b)
-10-
-------
The integral in (8b) is k~own as the exponential integral and
25 .. 16
has been extensively tabulated Juntgen and co-workers
have used the approximation,
T E
f enclT~
o
R-'- - %r
Ll..e
e
( 9)
which is satisfactory so long as the quantity ~ T is much less
than unity. Substituting ( 3), (9) and the relation [A 10 = ~V 0
into (8b) gives
\. £]
- !!!. RT e - -'1
\To-v = '\l.exp~ M e
( lOa)
If we now differentiate (lOa) with respect to temperature we
obtain
E }
- dV - ~, i. {eJCO r- ~ Ill'\.e - RT]
IT - V. ~T ' [M E
( lOb)
-
E .;
dV - t7 !'R.e)Cpr_~ 1!"'e-iTl! 'Te-RT)
iJT - V. M E [t1 E .aT \
( 10 c)
... -
.tV - ", !!, E/£ +IT'e-ir ex, I- ~ !!\.e-iT]
-;FT- Vo M E "R I) 't: M e
( IOd)
-11-
-------
b R .
which upon multiplication through the bracket term y E glves
E E
d.V - \T !!, fl'" ~'\:'trTeXD L ~ ~'\.e ii'T]
'"IT .. Vo M ~ err [ M e
( 10e)
RT
Since the approximation (9) is valid only for E < < 1 we have
finally the result
J.V
n=
v~. exp L- ( ~
£.
RT'- - RT
-e
e
+ fT) J
(11)
for the case of first order decomposition reactions.
Expressions
for other reaction orders have been derived by Van Reek, Juntgen
16
and Peters.
In order to relate the volume of gas evolved in a given
experiment to a fixed quantity of solid, the terms V and V 0 in
(11) are taken as volume of gas evolved per unit mass of solid
sample. Further, the experimental quantity actually
measured is the fraction of the total flowing gas that is the gas
evolved by the decomposition.
If we let P be the fraction of the
total flowing gas (carrier gas and evolved gases) that is the ev-
olved gas, G be the weight of the solid sample. and Q be the
volume flow rate of carrier gas, then the left hand side of ( 11)
is evaluated from the experimental variables by the relation
-12-
-------
dV -
-
dT-
?Q
-
MG
( 12)
so that an experimental value for :~ in cm 31 g-deg. can be obtained
at each value of T and M. From experimental knowledge of T and
M we can evaluate the kinetic parameters ko and E which are charac.
teristic of a given solid sample.
A typical gas production curve, by which we mean a plot of
dV
dT versus T, for a given M, is shown in Fi gure 1a for methyl
mercaptan evolution at a heating rate of 5° Imin. Actually, in
the experiment, of course, the experimental gas production curve
is obtained from measurement of ~~ as given by (12) as a function
of T. At the maximum in the curve :T (:~) =0 and it can be easily
shown 16 that solution of this equation yields the following expres-
sions fo.r the kinetic parameters ko and E
E=
'L ~RTQ
eRTo I~V" e- f
~ l;JT ITa
( 13)
ko=
ME
-e
~To
E
-
~To
( 14)
-13-
-------
100
80
~
)....
'"
~
~
~ 1: '0
~~
~
~
~
-4-0
20
300
.lfoo roo
7EM PCRATLlRJ: (oc)
la.Evolution of CH3SH in a non-isothermal pyrolysis in hydrogen.
_.- experin1.ental data
;( calculated from values of E and ko by analysis of
experimental results.
Figure
-14-
-------
Obviously, the transcendental equation, (13), cannot be
solved explicitly for E. However, we can solve it graphically
to obtain from a given experimental :~ vs T curve a numerical
value for the activation energy E. Thus, if we define the dimen.
sionless parameters
a=
e
-
RT~
( 15)
~:
'0 (J.V)
~ 'IT Tb
( 16)
and substitute into equation (13) we may write
(1- i)
ol:~e.
( 17)
Taking logarithms we have
~
-
0(
+~cX
-
..
H~~
( 18)
Th.e quantity A + lnoC. is plotted as a function of 0( in Figure Ib
Thus from this figure we may obtain a unique value for oi if we
2
know the value of the function, oc. + lno(.
-15-
-------
3.4
3.3
3.2
rn~ +0/0(
1+ln~(~~)TJ 3. 1
I
....
0' 1+ In"
I
3.0
2.8
14
2.9
16
20
22
24
26
18
oc. =E/RTo
. Figure lb. Graph for determining the activation energy for first order
rea.ction $ frorY1 the e='Cperir:n.en.ta..l p~ra.rn..e1:.er 6. de£i:ned i,....
28
-------
The procedure to obtain a unique value of E from the experi-
mental ~~ versus T curve is first to calculate a value ofP according
to equation ( 16), in. which ~ is defined in terms of the experimental
parameters V 0' To and C ~)-ro' From this value of e- we may cal.
culate 1 +ln~and, hence, by virtue of equation (18) we have the
quantit~i + lnG(. From Figure 1 b and the computed value of the
ordinate ~ + lna(, we obtain 0( . From this value of oi, we calculate
E by means of equation (15). As soon as E is thus determined, ko
can be computed immediately from equation ( 14), or in terms of the
dimensionless parameter", ko is given by
k.,=
Met ot
-e
~
( 19)
A specific example of this method of evaluation of ko and E is
as follows :consider the evolution-rate curve for methyl mercaptan
shown in Figure 1a.
In this figure we have plotted the intensity of
the m/e 47 ion in millivolts as a function of the temperature for a
pyrolysi s in hydrogen at a heating rate of 5° Imino
The intensity of
this ion is proportional to the concentration of methyl mercaptan in
the flowing gas stream and by means of prior mass spectrometer
calibration may be easily converted to concentration; however,
such a conversion for the purpose of obtaining the kinetic parameters
from this data is not necessary, since these parameters are obtained
from the dimensionles s quantities and defined by equations ( 15) and
( 16). Thus it suffices simply to use consistent units throughout,)
which means that the volume of methyl mercaptan evolved (area
under the curve in Figure 1a) is obtained for these purposes in the
units of millivolt-degrees/gram.
-17-
-------
Thus from equation ( 12) , we see that dV is simply proportional
dT
to the concentration of methyl mercaptan in the gas stream or to the
peak height of m/ e 47 in millivolts.
From the curve in Figure la we
obtain the following values:
Vo: '.0 . It} ,";I/'~J'.J"",..;',.tM.
To: 71#0.'<.
(~) = ~O ...~U,'".lf/"...
alT To
From these experimental values we calculate ~(eqUation 16) as
~=
"I!.(L~= q."
Vo \:dT
.
and hence
A~=
a.~'
Then by equation 18 we have
H.e....~ = 3.21 = ! + L~
With this value for the quantity 2/- "'lno( we now may use Figure lb
(as shown by dotted lines on this figure) to calculate at as
oi':. ~&/.'
Now by equation (15) we have
E = o('RTo =
3' ~t.o' IM.'~
-18-
-------
and by equation ( 19)
~o ~
Mot
T.
et
e
-
-
,
i.ox./IJ
"'1
"'" 11\
The dotted line in Figure 1a is calculated from the theory using the
above kinetic parameters E and ko and the heating rate of 5° Imino
agreement is seen to be quite satisfactory 0
The
2. PYROLYSIS IN A REACTIVE ATMOSPHERE WITH NO BACK REACTION
In the pyrolysis of coal in a reactive atmosphere we are concerned
with the reaction of a solid with the reactive gas to form a gaseous pro-
duct .
For the overall reaction in this case we replace (1) by (20) viz:
AsQ,,~ \- 1>8' ( ~ ~s.I"1 + C 3-"
( 20)
In a reaction such as (20) the rate is often first order with respect
to both reactants and for convenience and simplicity we will consider
that to be the case here 0
Then instead of ( 2), (3) and (4) we have
( 21), (22) and (23) viz:
-19-
-------
dtAl
- '"(f'!t =
,,~ LA 1 (3))
( 21)
tAl
-
-
o«(~-v)
( 22)
,Iv
(,l1=
R~ t]) 1 (~ -v)
( 23)
where kr is the rate constant pertaining to the reactive atmospherel
[Dl is the concentration of reactive gas in the atmosphere, and all
other quantities are as described for the case of the inert atmos-
phere.
For any given experiment the concentration of reactive gas in
the pyrolytic atmosphere is maintained constant so that we may in-
corporate this constancy into the second-order rate constant kr. and
write
~'=
R~lb]
( 24)
where kl is a pseudo-first-order rate constant. Hence kl has the
same units as k in equation (2), namely, seconds-I, but its numer-
ical value depends upon the concentration of reactive gas in the
pyrolytic atmosphere.
-20-
-------
In view of equation (24) we may express (21) and (23) as
below:
-~: \q'lAl
( 25)
dV
--
114; -
R' ('l.- V)
( 26)
Since (25) and (26) are mathematically the same as (2) and (4) ,
the remainder of the treatment for this case of a reactive atmos-
phere is identical to that of the inert atmosphere with k and ko
of the latter treatment being replaced here by k1 and kb If we
wish to evaluate the second order constants kr and kr, 0 we need
only divide kl and kol by the concentration of the reactant gas.
3. PYROLYSIS IN AN INERT ATMOSPHERE WITH BACK REACTION
For cases in which the back reaction cannot be neglected the
modification of the theoretical treatment is more involved.
For
this situation we write the chemical reaction as
As.,..,
~ )
( It..
B Sol.'J + C3&1
( 27)
-21-
-------
where kb is the rate constant of the back reaction.
In the kinetic model for this complex situation it is necessary
to include explicitly the geometry of the solid reactant and the velocity
of flow of the inert sweep gas.
For the present application we are
concerned with a static bed of reactant of cross-section A and depth
d, so that for a volume flow-rate of sweep gas of Q (cm3/min) the
average residence time, t, of a volume element of gas within the bed
is given by
f:
Ad -
q--
VB
Q-=
J
-
-.r
( 28)
In ( 28), VB=Ad is the volume of the reactant bed and v is the average
linear flow rate of the flush gas,
The rate of change in the concentration of the product gas C
with residence time 1" in the bed is given by
dt~l
d't ::
~ LAl- ~~Lc1
( 29)
where the first term, k[ A] , is the rate of formation of C and the
second term, kb[ C] , is the rate of disappearance of C due to the
back reaction~which we assume, for mathematical convenience,
may be taken as first order overall. The term [A] is the instan-
taneous concentration of solid reactant A and is assumed to be uni-
form over the bed.
-22-
-------
Integration of (29) over the residence time~1; with the
boundary condition that [ C] = 0 for1'=- O. gives
[cJ,.. =
R ~ - tttal)- }
- [A1 ,- e.
Rb
( 30)
For a non-isothermal experiment. the rate constants k and
kb. being functions of tempt::rature. are also time-dependent.
However, for moderate heating rates and reasonably short resi-
dence times. the change in temperature (AT:. t1t)iS sufficiently
small that the variation in the rate constants during the time "..
may be neglected. The concentrations of solid reactant A and
product gas C, namely, [A] and [ C] J are functions of real time,
t, (just as k and kb are) since the sweeping-out of the product gas
C depletes the reactant.
Explicitly this is given by
d[A1
di~-
Q tc11-
-
~ -
- t C 11'
1-
( 31)
Combination of (30) and (31) yields (32) ,
viz.
.I
.L tAtA'
[A1 dt
I
-
= - If'"
lot l~)
"\a(~)
f,-
e It..(t)t J
( 32)
-23-
-------
where we denote explicitly that the rate constants are functions of
real time t.
Now for the constant rate-of-heating experiment,
time and temperature are related by
di =
.L tiT
M
( 33)
so that substitution of (33) into (32) and integration gives
~
[AlL -
[A1o -
.L rT ~(T) r 1- i 1a..(T)'t I cLT
- M t ) te~ (T) l )
o
( 34)
where [A] 0 is the initial concentration of reactant [A] in
the bed. The same considerations of the lower limit to this inte-
gral as presented in the development of equations (8a) and ( 8b)
apply here also.
By using a series expansion of the exponential,
this integral can be evaluated numerically using tables of the expo-
nential integral or it may be evaluated analytically with the aid of
the approximation given in equation ( 9). The results, however,
are cumbersome and not very enlightening.
The case which is most pertinent to the present experimental
study is that of a very fast back reaction, or, in other words, a very
large kb.
For this situation we may assume that
k ~ » I
b
( 35)
-24-
-------
over the temperature range of interest.
With the aid of ( 35) ,
(34) reduces to ( 36),
viz:
in
[A],
[A10:= -
1.. (~(T) dT
Mt ~ "~("\
o
( 36)
Using the Arrhenius expression for the rate constanta, namely
E
k(T) ~ \~o e- RT
( 37)
-~
RblT):: ~bo e RT
( 38)
equation (36) become s
~T (E-E.)
A [AJT = - ..L J!.t.. e- RT dT
"IfM, [;4 J 0 t1 t ~b. 0
( 39)
Employing again the approximation (9) for the exponential inte-
gral we find from ( 39)
{G..e..)
f- J... (tt.. \!''' - ~T ]
fA IT -:. [Alo elCp[ M'i" "fi"...tj:i,.> e (40)
-25-
-------
Differentiation of (40) with respect to temperature yields the gas
production rate (related to ~ by means of equation ( 3)) as
-t -~-E"'~
dV cJiAl tA1.~. ex r f(~f..' J. ~ It e. RT 1 (41)
- 0( dT: 7!f = - M1=~. Pc t~J'" ttt hbolt-E~)
For the case of an inert sweep gas with no back reaction, we showed
earlier that equation ( 11) applies, which with slight rearrangement
in accord with (3) may be written
\ If 1
dv do Al A' - 1. --
- at r: ~: - ~ L AXP - ( £. + 'f1t.R1 e raT}
".1 liT Hfa "iF t 1t1 Hte
( 11 b)
Comparison of (41) with ( 11 b) shows that the depe'ndence on
temperature is identic.a) with E-Eb in (41) replacing E in ( 11 b).
Hence the same graphical techniques described for the case of no
back reaction, (equation (13) -( 19)), may be applied in this case,
When this is done the "activation energy" obtained is the difference
(E-Eb). The pre-exponential factor is obtained as (~ +-)
d. t. k bt' d" 1 h . ko b.
an SInce IS nown we 0 aln lrect y t e rati0"k'Eb .
4. PYROLYSIS IN A REACTIVE ATMOSPHERE WITH BACK REACTION
In this situation we have the reaction
Ak'"''
+ °5'"
R¥
)
-------
As for the case of a reactive atmosphere with no back reaction we
assum.e the forward rate to be first..order with respect to both re-
actants and hence second.order overall
Thus)for this case" kr
is the second.order rate constant pertaining to the reactive atmos-
phere (cf (21)) .
Again) for a given experiment the concentration
of reactive gas is maintained constant so that we may incorporate
this constancy into the second-order rate constant kr)as in ( 24) .
The treatment for this reactive atmosphere with back reaction is
then identical to the treatment described in equations (28) -( 41)
with the exception that kl:kr [D] replaces k.
Hence from a study of the gas evolutbn curves" using the treat-
ments described we may in this case obtain Er-Eb and ~ '= ~ .
I ~o ~
Since [Dl is a known quantity, we thus obtain It,... ~.
hbo
5. NON-ISOTHERMAL KINETICS OF THE BACK REACTIONS IN COAL
PYROL YSIS
In our treatment of Case 3, viz'l
t?
A,ot.a
»
<'h
b
T3'611~ oJ C "'"
( 27)
we showed that an analysis of the gas evolution curves of C leads to
values for the activation energy difference E-Eb and the frequency
factor ratio .!!a
"~.'
Similarly the same analysis applied to Case 4,
viz;
A.sc.,. ~ ... 1)~" ~
,,~
..
.. ",-
13$~I. ~ + C "...
( 42)
-27-
-------
permits evaluation of the activation energy difference Er - Eb
~Q
and the frequency factor ratio ""0' In neither case can the
kinetic parameters of either the forward or the reverse reactions
be obtained independently as long as conditions are such that
both reactions are occurring to significant extents.
Two methods that may be used to obtain the kinetic parameters
of the forward and reverse reactions independently are suggested by
consideration of equation ( 32), viz.
d[~'
---
tAt -
d [(.1
--
.... -
" ( 4)
b..(ot.)
( - I,..
-------
Suppose a coal sample is first desulfurized to completion and
all gases removed.
In terms of the notation used, we have now a
sample of Bsolid (coke), the Asolid being completely converted.
We may now introduce Cgas (HzS), in either an inert sweep gas
or a reactive sweep gas, Dgas (Hz) , to the sample of Bsolid ( coke)
and study either the evolution of Dgas or the consumption of C, via
the reaction
c~., ... ~$."CI
~b
..
As.,:" + 1)3'"
( 44)
The amount of conversion of Bsolid to Asolid is kept small so that
only the reaction (44) is studied. With the assumption that (44) is
first order in [ C] ~ [HzS]).I and that [B] is so large that it may be taken
as constant over the reaction, we have
d tel
tJ t :
- Itb tel
( 45)
where [cl is the concentration of HzS, and as before
"b:
- t:b
k e RT
b,.
( 38)
-29-
-------
For a given temperature of the bed we may integrate (45) to obtain
the concentration of HzS in the gas leaving the bed, viz:
[c 1, =
[el. e- "..1"
( 46)
where [ c] 0 is the initial concentration of HzS is the inlet gas and 1-
is the average residence time in the bed.
For a non-isothermal
experiment in which the heating rate is sufficiently low that the
temperature change during the residence time 1-is small, kb in
(46) may be considered as a constant.
Assuming that this condition
is met we may substitute (38) into (46) and obtain
A feJ - - kb 1-
",. [e'. - .
E
_.EJL
RT
e
(47)
Taking logarithms of both sides of (47) and converting to base 10
gives
~J (L Wt) =
J. Ik t) E. ( J.)
~~ bCII -~JolR T "
( 48)
Hence, by plotting semi-logarithmically the quantity In CC'J 0
1 [c.1
versus T' the kinetic parameters Eb and kbo"t may be determined
from the slope and intercept, respectively. Since 'fis known from
the flow rates and bed size, kbo is determined.
-30-
-------
Instead of measurement of the decrease in C (HzS) due to the
J J
back reaction (44), we may equivalently measure the formation of
D (Hz) .
,
from ( 44)
Fora 1: 1 stoichiometry, which seems reasonable, we have
((.1- ~ r"l. - tt) 1
'\
( 49)
so that (48) becomes
POJ(t,. :~~:tJ)V: .Q~(~b:t) -.1~ (~)
( 50)
and the kinetic parameters are obtained from the slope and intercept
te'1 ~ J [.,""0
of a semi-logarithmic plot of In =... [ J ttJ-'
{'']Q - [J)1 N,,' 0 - ~
.L.
. -uC T '
,
,
-31-
-------
III.
EXPERIMENTAL METHODS
Two types of experiment were carried out in this examination of
coal pyrolysis. The early experiments were conducted under iso-
thermal conditions with a flush gas (-either helium or hydrogen) flow-
ing continuously over a sample of coal held at a fixed pre-determined
temperature. Fast heating experiments were performed by dropping
the coal sample on to a quartz wool bed in the reactor preheated to the
desired reaction temperature.
The volatile products of pyrolysi~ J swept
out of the solid sample by the flush gas, were collected in cold traps and
subsequently analyzed by a combination of gas chromatography and plasma
spectroscopy. Tar and coke produced were analyzed for sulfur by the
ASTM procedures.
The second type of experiment carried out, and the most informative,
involved continuous mass spectrometric measurements of products pro-
duced under non-isothermal conditions. In this technique the coal sample
is heated at a linear and known rate while the flush gas (again helium or
hydrogen) continuously sweeps through the coal sample and removes vol-
atile products.
A portion of the effluent gas is fed to a mass spectrometer,
which may be tuned to monitor continuously with time a pre-determined
ionic mass or to scan repetitively a selected region of the mass spectrum.
Most of these experiments were conducted by scanning over the range
from mass 1 to mass 84 with a period of one minute. Since the heating
rate was usually SOC/min this mode of operation yields an experimental
point on the gas evolution curve as a function of temperature every SoC.
For all of the sulfur compounds and for most of the other major compounds
of interest, it was possible to identify a single representative ion in the
spectrum which was relatively free from interference by ions produced
from other substances present. With the mass spectrometer operating
-32-
-------
at constant sensitivity the amplitude of the signal (usually expressed in
millivolts) corresponding in the mas s spectrwn to the ion representa-
tive of a particular compound is proportional to the concentration of that
compound in the effluent gas.
A schematic diagram of the apparatus is
shown in Figure 2, and a detailed description of the apparatus and pro-
cedures is given in the Appendix. With this arrangement the gases
evolved during the pyrolysis of coal are monitored quasi-continuously
as a function of time, and through the linear relationship between tem-
perature and time also as a function of instantaneous temperature.
The
theoretical bases for obtaining kinetic parametersfrom the data obtained
in these non-isothermal experiments are given in full in Chapter II.
In both types of experiments the gases condensable at liquid nitrogen
temperature (including all of the sulfur compounds) were trapped from
the reactor'. effluent and quantitatively measured using the gas chromato-
graph.
These ~ asurements coupled with the tar and coke sulfur analysis
by the ASTM procedures were used to obtain sulfur balance determinations
for each experiment.
The coal sample, char residue, and tar from each
experiment were weighed in the analytical balance.
The total gas produc-
tions were determined by difference from these results, since many of
the most abundant gases, e. g. Hz, CH4, CO, are not effectively trapped
at liquid nitrogen temperatures.
For some of the experiments the cal-
ibrate d mass spectrometer was also used for quantit tive gas evolution
measurements.
Comparison of the GC and MS results were generally
in good agreement.
In the non-isothermal kinetic measurements only
the relative yield of a particular gas as a function of temperature is
required in the calculations; therefore, in many of the experiments the
conversion to absolute units was not made.
-33-
1':":
-------
:RrAcm
I
W
~
I
FJ.t/SH
G-A$
HE OR
Hz.
SA"'''' J. E"
F"V1!lNAC.e
CDNrRtJj.
JJ"'~A~
T6".M~"'T"It£"
~O~..~~
1"'1"'1
1'rlll.MEr6R
HOT Zo#c-
7iI(!!"RMQl'OVPJ.. ~
NI"~ZI~I!"
Y...~Y€
]A.7t
TR~P
L'~IIUJ #2
7i-AJItS FO~ t; .c.A"'A~ "t'.$'fF'S"
:1".::.;'&"""""- 2.. Sc;;.b.er>;>.a.ti.c: ""i.a..""=-~ co'" -~~-=-'=>.._'"'t...' ~~~~=_tu..."
ff""PFRATVllc
TiME
'"RF J.." rlJlt-
C ON~~"'rRAr/(Jif
,tt.u.s
S7'~t:'TI(O
71 ".~
V£Nr
-------
The temperature measurements in these experiments were
accomplished usi ng a Chromel-Alumel thermocouple attached to the
furnace block immediately adjacent to the position of the reactant bed
in the reactor.
The
annular space between the reactor and the furnace
was tightly plugged with quartz wool both top and bottom to prevent air
convection through this space.
The furnace used in this work gives a
hot zone 12" long and the reactant bed (typically 1" long) was placed
at the center of this zone.
In the experiments involving high flow rates
of flush gas, quartz wool was packed into the reactor on the inlet side of
the reactant to provide better thermal contact wi th the gas.
Comparison
measurements between the temperature indicated by the external thermo-
couple and one installed in the reactant bed gave agreement within .:.lOoC
over the range of experimental conditions used in these studies.
details of the experimental procedures are given in the Appendix.
Additional
1.
TYPE OF COAL
Most of the experiments in this work were conducted on a No.6 type,
high volatile, rank C, Illinois coal containing approximately 5% sulfur.
However, non-isothermal experiments were also conducted on an Illinois
coal containing approximately 1% sulfur.
coals are given in Table 1.
Typical analytical data on these
-35-
-------
TABLE 1. SUMMARY OF ANALYTICAL DATA ON 5%
AND 1% SULFUR COAL
Illinois Coal - 5%
Illinois
Coal-
1%
Forms of Sulfur
(per cent of coal)
Sulfa te
Pyritic
Organic (diff)
Total
0.31
2.50
1. 95
4.75
0.01
0.26
0.60
0.87
Proximate Analysis
(per cent of coal)
Moisture
Volatiles
Fixed Carbon
Ash
Sulfur in Ash
(per cent of ash)
Sulfur in Fixed Carbon
(per cent of coke residue
from volatile matter
dete rmina tion)
9.7 8.7
34.0 34.8
41.7 50.0
14.6 6.5
0.02 0.51
3.75 0.60
Ultimate Analysis -
Car-bon
Hydrogen
Nitrogen
Sulfur
Ash
Oxygen ( diff)
dry basis
64.7
6.0
1.6
5.5
16.2
6.1
77.5
5.7
1.9
0.94
6.6
7.3
-36-
-------
IV.
RESULTS AND DISCUSSION
1.
GENERAL NA TURE OF COAL PYROLYSIS
Isothermal Exper iments
The set of isothermal experiments in which the evolved gas com-
pos ition and the extent of desulfurization of the solid reactant were
studied as functions of temperature, flush gas (He or Hz) , heating
rate, reaction time, and coal particle size have confirmed the gen-
eral picture provided years ago by the pioneering work of Powe1l2a, b
3
and Snow.
The results of these isothermal pyrolyses are summarized
in Tables II and III.
With regard to sulfur-removal efficiency from the solid reactant
by reaction with hydrogen, pertinent results of our isothermal experi-
ments are compared in Figure 3 with the corresponding values obtained
3
36 years ago by Snow.
The agreement is quite good and confirms the
general picture of 1932 that sulfur removal by some process begins at
approximatel y 300 ° C, the amount removable by this proces s reaching
a limit at about 500 ° C, and that other removal processes begin to occur
above about 700°C.
The results for isothermal reactions in an inert
gas (helium) summarized in Table III are also in good general agree-
ment with the earlier work by Snow using nitrogen as the inert gas.
-37-
, iJ
-------
TABLE H. SUMMARY OF DATA FOR ISOTHERMAL HYDROGEN RUNS
Sulfur Recovery, %
Coke Desulfuri- S02
Run Coal Temp. Residence Time Yield zation CH 3SH Total
No. %S °C Reaction Time % Factor Coke Tar H2S CS2 C4H4S Re covery
13 5.4 700 3.0 sec. 50.9 1.2 33.5 3.7 35.4 <0.1 < 0.8 73
10 min.
14 4.6 1000 3.0 sec. 53.0 0.71 39.8 0.2 25.4 2.5 < 0.4 68
15 min.
15 4.6 700 3.0 sec. 52.2 1.1 40.1 4.4 37.9 < 0.1 < 3.7 82
15m in.
16 4.6 700 3.0 sec. 52.2 1.3 34.6 5.2 40.3 < 0.1 < 0.7 80
15 min.
I
w 17 5.0 1000 3.0 sec. 52.4 1.5 32.0 0.1 37.4 11.6 < 0.6 81
ex>
I 15m in .
18 4.7 1000 3.0 sec. 50.9 1.2 45.8 2.1 43.9 9.6 <,0.2 101
15 min.
112 5.0 1000 3.1 sec. 52.4 1.4 38.9 1.8 39.5 11.7 <,0.4 92
15 min.
"---
113 5.0 1000 2.8sec. 50.5 3.2 21.8 2.8 57.6 8.9 < 0.4 91
4-1/4 hr.
114 5.0 1000 11 sec. 50.5 1.1 47.6 2.5 37.6 12.4 < 0.4 100(6)
4-1/4 hr.
115 5.0 1000 3.5sec. 53.8 3.6 24.2 7.6 55.9 0.3 1.3 89
4-1/4 hr.
116 5.0 1000 2.7sec. 52.2 7.1 10.7 6.5 69.6 0..9 0.7 88
4-1/4 hr.
117 5.0 1100 3.0 sec. 50.1 3.8 16.8 0.5 55.6 9.7 < O. 2 83
4-1/4 hr.
-------
( 1)
( Z)
( 3)
( 4)
( 5)
I (6)
"'"
..0
I
( 7)
Notes For Table II
all 3 gram samples except 114 which was 11 grams.
all samples ZO /40 mesh coal except 115 which was 100/ZOO mesh.
quartz reactor for all runs except 13, 14, and IS which were in stainless steel reactor.
all coal injections into the furnace were gravity feed at the indicated temperature except US
and 116 in which the coal was in place prior to heating.
purge gas at 1 atmosphere pressure.
assumed efficiency because a portion of the gas sample was lost during analysis - other
sulfur recovery values are based on this assumption.
In 115 and 116 the rate of heating was approximately 33° Imin.
-------
TABLE III. SUMMARY OF DATA FOR ISOTHERMAL HELIUM RUNS
Sulfur Recovery, %
Coke De sulfur i- SOz
Run Coal Temp. Res idence Time Yield zation CH3SH Total
No. %S °C Reaction Time % Factor Coke Tar HzS CSz C4H4S Recovery
11 3.8 700 1.0 sec. 55.8 0.74 54.2 4.4 36.1 < 0.1 < 0.1 95
10 min.
12 4.0 1000 3.0 sec. 59.2 0.75 52.4 4.2 27.4 7.5 < 0.3 92
10 min.
19 4.8 1000 3.0 sec. 53.9 0.93 46.2 1.3 30.9 10.3 < 0.5 89
15 min.
110 5.0 700 3.0 sec. 56.5 0.94 46.2 5.0 38.6 < 0.1 < 0.6 90
15 min.
I III 5.0 700 3.1 sec. 55.0 0.85 48.3 5.4 36.3 < 0.1 1.4 91
>+>-
a 15 min.
I
118 5.0 1100 3.1sec. 54.8 1.8 34.8 0.4 55.2 6.3 < 0.1 97
4-1/4 hr.
119 5.0 900 23 sec. 58.3 0.91 49.5 6.3 37.5 < 0.2 1.4 95
1.7 hr.
Note s
(1)
all 3 gram samples except 119 which was 23 grams and 11 which was a combined total of three 1 gram
pyrolyses.
all samples 20/40 mesh coal.
( 2)
( 3)
( 4)
quartz reactor for all runs.
all coal injections into the furnace were gravity feed at the indicated temperature except 119 in which the
coal was in place prior to heating.
( 5)
( 6)
purge gas at 1 atmosphere pressure.
in 119 the rate of heating "Was approxiITI.ate1y 33° /rriin.
-------
IV
9
e
~ 7 []
~
j5 ~
II) ~ 6
+ <:)
~ I,J
< V)
e"
~ 5
~
<::>
.... + S.L ow H£ATlN8
~ Qb
~
~ 3 ~
~
0
~ G)
~ 2 FAST tlcATING
(.If HR. Re-ACrfON7iN4£)
---I
L fA:'T HEATING
(If Milt R~ACrICNTiAII~)
o
300 400 500 600 700 800 900 1000 /'00
T
DC.
Figure 3. Sulfur removal factors for coal pyrolysis in hydrogen at
one atmosphere.
@,O R.D. Snow, Ind. Eng. Chem. 24, 903 (1932).
X this work, fast heating; 15 minute reaction time
~ this work, fast heating; 4 hour reaction time
o this work, slow heating ( 33° /min)
-41-
-------
While useful in filling a confirmatory role of this old general picture
of coal pyrolysis, the isothermal type of coal pyrolysis experiment is
not well-!;,1.lited toward making further me chanistic advancef: in our
understanding of this complex chemical conversion and hence warrants
little further dis cus s ion.
However, a few results from the isothermal experiments should
be noted.
In the fast heating experiments to 10000 C in both hydrogen
and helium, Runs 12,14,17,18,19,112,113,114,117 and 118)a sub-
stantial amount of the total gaseous sulfur evolved is in the
form of carbon disulfide, while in the experiments employing the
slower heating rates, Runs 115, 116, 119,the CSz evolution is quite small.
Also the slower heating rate produces a substantial increase in the de-
sulfurization efficiency as previously observed by Snow3. All of these
isothermal experiments were conducted using coal in the 20/40 mesh
range of particle size except for run 115 which used a 100/200 mesh
cut.
This run, which was otherwise similar to run 116, gave, con-
trary to expectations, a lower desulfurization factor than obtained for
the larger particle size.
However, upon examination of the solid re-
s idue from run 115 after completion of the experiment, it was found
that the fine coal had agglomerated into a dense solid mass.
Therefore,
it appears that the effective particle size for this experiment, at least
during the high temperature part of the experiment, was in fact larger
rather than smaller.
-42-
-------
Non-isothermal Experiments
In the non- isothermal pyrolysis studies, the theoretical basis of
which was developed in Chapter II, the primary data obtained are the
rates of evolution of the various gases
dVi = M dVi
dt dT
( 51)
as functions of the temperature, where M is the heating rate of the
solid coal sample.
Such data which provide immediately a very
informative picture of the overall pyrolysis in hydrogen and helium
are shown in Figures 4 and 5, respectively.
The composition of the gas evolved is quite complex with Hz, CH4,
HzS, CzH6' C3Ha, C6H6' CO, COz, CH3SH, CSz, and SOz having been
firmly identified in relative abundances that are very sensitive to the
reaction temperature.
This variation of gaseous product composi-
tion with temperature is not unexpected since such a wide variety of
products are undoubtedly formed by a set of parallel and consecutive
chemical reactions, each proceeding at a temperature-dependent rate
in accord with its frequency factor and activation energy.
It is seen that with both flush gases the maximum rate of total
gas evolution occurs at 425°C, coincident with the maximum evolution-
rate of ethane and propane (and other hydrocarbons at lower concentra-
tions, not shown in the figures). Below 500°C the evolution rate of
-43-
-------
£i( cm3)
dT °C-9
I
>l>-
*""
I
Figure 4.
0.3
H2S
CH4
0.2
0./
502
o
100
zoo
300
1000
110
400
500
TEMPERATURE
600
(0 C)
700
800
900
Gas evolution curve for 50/0 sulfur coal pyrolyzed in hydrogen showing
containing species and major hydrocarbons, Run N3, seeTable IV for
experimental. detaUs.
sulfur
-------
dV cm3
'iT (OC-, )
I
~
U1
I
0.30
0.20
0.10
o
100
Figure
H2 x 0.36
,,,,,...,
I ,
. ,
I I
: I
.
,
,
,
,
,
,
,
I
I
I
I
,
1
I
I
I
/
502
200
300
400
500
600
700
800
900
1000
1100
TEMPERATURE
(0 C)
5.
G"',e,'>.volution curve for 5% sulfur coal pyrolyzed in heliurn showing
sulfur containing species and major hydrocarbonc:, Rup N4, see
Table V for experirLelltal ;ie.ail:::.
-------
methane is comparable with both flush gases but above 500°C the rate
of methane formation is significantly greater in the hydrogen atmosphere.
Hydrogen evolution from the coal was not measurable in the hydrogen-
atmosphere experiments but in the case of a helium atmosphere. for
temperatures above 550°C. it proceeds at a higher rate than that of
any other gaseous product. Moreover. the area under the evolution-
rate curve shows hydrogen to be the most abundant product of a complete
coal- pyrol ys is .
While interesting and complex problems in their own
rightJthe evolution- rates and evoluti on-me chanisms of the hydrogen and
hydrocarbons are not of prime interest in this research and. except
where they bear directly upon the desulfurization kinetics, will not be
dis cus s ed further.
All of the sulfur- containing species show maxima in their evolution
rates below the maximum rate of total gas evolution, namely, 425°C.
The maximum rate of evolution of sulfur dioxide is at 300 ° C with both
hydrogen and helium flush gases.
The maximum evolution-rate of
methyl mercaptan and the first maximum (with increasing temperature)
in the evolution rate of hydrogen sulfide occq.r at 375°C in a hydrogen
atmosphere but are shifted to 400°C when helium is used as a flush
gas. The second maximum in the hydrogen sulfide evolution-rate that
is observed in the helium atmosphere does not correspond to any peak
found in the hydrogen atmosphere. but interestingly does appear to coin-
cide with the onset of hydrogen production in the helium atmosphere.
This coincidence suggests that the second maximum found in helium
may be due to a reaction of sulfur in the coal with hydrogen produced
-46-
-------
from the coal; this second maximum. would not, of course, be observed
in the hydrogen atmosphere because of the constant presence throughout
the experiment of large amounts of hydrogen. For convenience, the
hydrogen sulfide evolution curves for the two flush gases are shown
together in Figure 6. Here we see that the low-temperature maximum
in helium is lower in amplitude and lies at a temperature 25° higher than
the corresponding maximum. in hydrogen.
Such a shift would be expected
if a rate process similar to that described in Section .II-2 viz;
('x- S )'0&1
...
" "
\41
..
\..\ 2.. ~
4- )(,.1:1
($1)
was involved in both atmospheres. The term "Hz" devotes hydrogen
in close proximity to sulfur in the coal and ( since it is supplied externally)
would be greater in the case of the hydrogen atmosphere than is the case
of the helium atmosphere. To show that the shift is in the direction
predi cted theoretically. consider equation (11) as modified for the case
of a reactive atmosphere with no back reaction as described in Chapter II-2.
Thus for HzS evolution we have
d%,s -
tAT -
v.lt~
M
[ -' ,,'
ex, - (fT ... ~
.'
!r" e - ~T]
E'
(51)
-47-
-------
.30
.zo
I
~
00
1
~ cm3
err (og-g)
.10
-~""",,~."-'~-~-.' r-~.'~ ...-- ...._,_.~.'--------- ---~---_._._.- -----. ---. ------------.--. ...
HZS Produced In HZ
o
100
Figure 6.
,.,
I \
I ,
I ,
J. \
I \
I \
I \
I \
\
\
\
/-,
I \
I \
I I
\
I "\
I \
1\/
I \ /
/ '_.I
/
/
_/
200
HZS Produced In He
300
400
500
--~---~--
600
800
900
TEMPERATURE
"1000
1100
.
700
(0 C)
Comparison of HzS evolution in Hz atmosphere, Run N3, with HzS
evolution in He atmosphere, Run N4.
-------
where ko1:kr [Hz]. The variation of this rate with [Hz]
pressed as the variation with the Pseudo fl"rst d
-or er rate
For a maximum l"n ,LVH-a.S
- we have then
doT
can be ex-
cons tant ko 1.
~ ( J. V~h')
~R~\7T = C)
( S't)
or
,-
..~
-
t1
'\.
R T ~AIt
e'
61
--
e RT"-'M
=0
( $)
This transcendental equation cannot be solved explicitly for the
dependence of Tmax on ko 1 (and hence on [Hz] ) .
However, for
a very small temperature shift from some reference temperature,
T 0' it is easily shown that
-49-
-------
t:1t~
e'
-
~ -
R IJtA~. ~ ')
lii If' ..
~'
1? A.(J:, . ... Cu.3)
( 51.)
showing that Tmax decreases as [Hz]
increases.
Such a result
would be predicted also for the case of a reactive gas with back
reaction (Section II-4). We may conclude, then, that the low-
temperature maximum in the hydrogen sulfide evoluti on-rate arises
from a reaction whose rate depends upon the concentration of hydrogen
in the vicinity of sulfur as well as on the sulfur concentration. The
reaction involved cannot be the intramolecular elimination of HzS from
compounds originally contained in the coal, unless such compounds were
first hydrogenated in the rate determining step.
Above 750°C, practically no additional hydrogen sulfide is evolved
in a helium atmosphere while in a hydrogen atmosphere a process (or
processes) occurs whose rate attains a broad maximum at 1000°.
The overall temperature-(or time) integrated results of the non-
isothermal pyrolysis experiments, which are in a form to be compared
with the results of the isothermal pyrolyses ~ are summarized in Tables
IV-VII.
The effect of hydrogen flow rate, or gas res idence time, was investiga.
ted in non-isothermal runs Nl, NIO, NIl, N12, and N23.
The effe ct of
helium flow rate was checked in runs N2 and N13.
The effe ct of particle
size on the desulfurization kinetics was studied in runs N5, N6, N7, N8
on samples of coke to avoid difficulties due to agglomeration. Most of
these experiments were conducted on a 5% sulfur Illinois coal, but some
-50-
-------
TABLE IV. SUMMARY OF COAL PYROLYSIS EXPERIMENTAL DATA FOR NON-ISOTHERMAL
HYDROGEN RUNS A T ONE ATMOSPHERE.
Run Coal Temp. 0C Residence Time Coke Desulfurization Coke Tar H2S CS2 S02' CH3 SH Total
No. %S Heating Rate Reaction Time Yield % Factor C4H4S Re cove r
Nl 5.0 25-1100 3.0 see 51.6 6.7 10.9 4.5 67. 1 0.3 < o. 7 83
5 ° Imin. 4. 0 hr
N3 5.0 125-1100 3.0 see 50.8 4.8 16. 1 4.6 73.0 < o. 2 1.7 95
6 ° Imin. 2. 7 hr
NI0 5.0 25-1100 0.72 see 51.2 9.6 8.5 5.6 67.0 < 0.1 9.3 90
5 ° I min 4 hr
NIl 5.0 25-1100 0.39 see 49.9 10.2 7.9 5.0 66.6 < 0.1 8.0 88
5 ° Imin 3. 9 hr
Nl2 5.0 25-1100 1. 7 see 52.4 9. 1 10.0 5. I 81 . 3 < O. I 3.4 100
I 5 ° Imin 3. 9 hr
l.11
.....
I N14 0.87 25 -1100 0.40 see 57.7 3.6 41.6 7.9 123 < 0.1 20.4 193
5 0 I min 4.0 hr
N16 4.2 25-1100 3.2 see 53.4 3.0 33.4 4.7 89. I < 0.1 6. I 133
5°/min 3. 9 hr
N 17 0.87 25-1100 3. I see 57.5 3.5 21.2 6.3 40.0 < 0.1 27.4 95
50 I min 3. 9 hr
N23 4.2 25-900 0.031 see 56.8 ( 5) < O. I O. 7 22. 8 < O. I 9.8 33(6)
5°/min 3 . 0 hr
-------
I
U1
N
I
Notes for Table IV
( 1)
all 3 gram samples except N23 whi ch was 0.308 gram
( 2)
,~ll samples 20/40 mesh coal
( 3)
quartz reactor for all runs except NI6 in which the stainless steel reactor was used and
N17 in which the stainles s steel with quartz liner reactor was used.
(4) \ n runs where the reaction time exceeds that required for the tempe rature rise at the given
rate, the temperature was held constant at the maximum temp. for approx. 20 minutes.
( 5)
since sulfur in the coke was not detected above the limits of sensitivity of 0.01 mg S, a
desulfurization factor cannot be calculated.
( 6)
sulfur recovery was low because gas trapping at the high flow rate used ( 1 l/mi n) was
ineffi dent.
-------
Run Coal
No. %S
N2 5.0
Temp ° C
Heating
Rate
25-1100
5°/min.
TABLE V. SUMMARY OF -COAL PYROLYSIS EXPERIMENTAL DA TA
FOR NON -ISOTHERMAL HELIUM RUNS.
Sulfur Recovery, %
Desulfuri-
Residence Time Coke za ti on
Reaction Time % Factor Coke Tar HzS CSz
3.1sec( 56.7 0.96 44.7 2.3 34.6 0.9
4.0 hr. 6)
SOz
CH3SH
C4H4S
Total
Recovery
4.4
87
N4
(
>
5.0
150-1100
5° Imino
3.0 sec.
3.2 hr.
53.3
(5)
41. 8 5.7 63.4 < 0.1 4.3
49.5 6.6 69.5 < 0.1 17.9
143
N13
57.3
5.0
I N15 0.87
\J1
w
I
Notes
( I)
( 2)
( 3)
( 4)
( 5)
( 6)
25-1100
5° Imino
25-1100
5° Imino
0.36 sec.
4.0 hr.
1.8
115
0.39 sec.
3.6 hr.
61.3
1.9
all 3 gram samples.
all samples 20/40 mesh coal.
quartz reactor for all runs.
purge gas at I atmosphere pressure.
coke sample lost during analysis, so these figures unavailable.
in cases where the reaction time exceeds that required for the temperature rise at the given rate, the
temperature was held constant at the maximwn temperature for approximately 20 minutes.
-------
Sulfur Recovery, %
Temp. °C Desulfuri- SOz
Run Pressure Heating Residence Time Coke za ti on CH3SH Total
No. atm. Rate Reaction Time % Factor Coke Tar HzS CSz C4H4S Recovery
j
N18 5 25-1100 J:5.5sec. 47.3 1.2 24.3 2.7 21.9 1.4 3.4 54
5° Imin. 3. 8 hr.
N19 5 25-1100 2.0 sec. 35.7 7.8 13.5 3.0 95.4 < O. 1 6. 1 118
5° 1m in. 3.4 hr.
N20 5 25-1100 3.7 sec. 40.1 5.1 18.0 4.5 82.5 < 0.1 4.4 109
, 33° Imin. 1. 8 hr.
U1 1 ( 5)
*'" N21 25-1100 0.37 56.0 1.9 26.6 3.8 46.4 < 0.1 2.0 79
I -
8 5° Imin. 3.4 hr.
Notes
( 5)
TABLE VI. SUMMARY OF COAL PYROLYSIS EXPERIMENTAL DATA FOR NON-
ISOTHERMAL HYDROGEN RUNS AT DIFFERENT PRESSURES
( 1)
( 2)
( 3)
( 4)
all 3.0 gram samples except N20 which is 6.0 gram.
all samples 20/40 mesh coal with 4.2% sulfur.
stainless steel reactor with quartz liner used for all runs except N21 in which quartz reactor was used.
in runs where the reaction time exceeds that required for the temperature rise at the given rate, the
temperature was held constant at the maximum temperature for that time.
purge gas was a hydrogen-helium mixture of 1 I 8 atmosphere hydrogen and 7/8 atmosphere helium.
-------
TABLE VII .SUMMARY OF COKE PYROLYSIS EXPERIMENTAL DATA FOR NON-
ISOTHERMAL HYDROGEN RUNS ON COKES OF DIFFERENT MESH SIZE.
Coke, Temp. °C Desulfuri- Sulfur Recovery, 0/0
Run o/ds Heating Residence Time Coke z'ation Total
No. Mesh Rate ReaCtion Time( 4) % Factor( 5) Coke HzS Recovery
N5 4.91 25-1100 1.5 sec. 89.2 4.9 27.9 53.7 82
20/40 4°/min. 4.4 hr.
N6 4.65 25-1100 1. 5 sec. 87.0 9.6 17.5 64.3 82
40/100 5° /min. 4.1 hr.
N7 3.76 25-1100 1.4sec. 82.5 9.1 22.6 76.1 99
100/200 5° Imin. 4.3 hr.
N8 3.72 25-1100 1. 5 sec. 73.1 10.4 20.3 78.6 99
minus 200 5° Imin. 4 hr.
I
U1
U1
I Notes
(1) all 1.5 gram samples.
(2) quartz reactor for all runs.
( 3)
( 4)
purge gas at 1 atmosphere pressure.
reaction time includes approximately 20 minute period at end of run when the sample was held constant
at the maximum temperature.
( 5)
desulfurization factor was calculated based on the original coal sample from which the coke was prepared,
see Run 119.
the CSz, SO z, CH3SH and C4H4S % S recovery was < 0.1 % in all runs.
( 6)
-------
comparison measurements were conducted on a 1% sulfur Illinois coal
in Runs N14, N15, and Nl? Analytical data for these coals are
summarized in Table I of Chapter III.
The effect of hydrogen pressure
was studied in runs N18, N19, N20, and N2l and the results are sum-
marized in Table VI.
These results indicate that increas ing the hydrogen pres sure from
1 atmosphere to 5 atmospheres, at constant gas res idence time, does
not significantly increase the desulfurization factor under the experi-
mental conditions employed.
Additional work on the effect of hydrogen
pressure on the reaction kinetics is required.
The results on the 1%
sulfur coal indicate that the desulfurization behavior of this coal is
qualitatively similar to that for the 5% sulfur coal.
Further work is
required to establish quantitative differences and to correlate these
differences with differences in amounts and forms of sulfur in the coals.
Investigations of the effect of particle size were carried out on
samples of coke, because of the tendency of the coal sample towards
agglomeration on heating.
The coke samples were prepared from the
ori ginal coal samples by heating to 9000 in helium at 1 atmosphere
pres sure and holding for 1 hour at this temperature.
The coke sample
so formed was ground and sieved into four approximately equal portions
of 20-40 mesh, 40-100 mesh, 100-200 mesh and minus 200 mesh.
Half of each sample was analyzed for total sulfur while the other half
was desulfurized in a non- isothermal experiment in a hydrogen
atmosphere. The results, summarized in Table VII and Figure?
indicate that below about 100 mesh there is little or no effect of particle
size, under the experimental conditions used in these experiments.
-56-
-------
~
IOO-ZOO MesH
2.()OM~S"
I
\JI
-oJ
I
~
c;)
.....
.~ 30
~
~
~-
~~
Q~
~':::) 20
V)~
~~
....
~iq
_lie:
...~ 10
<:
ij
Q::
I/O ~/()()Me.sH
:2.1> -~ ,., E S 11
.fDO
500
'00
700
800
900
lOCO
1100
r£MP£RATUR£ (OC)
Figure 7. Hydrogen sulfide evolution from different particle size cuts of coke by non- isothermal
reaction with H2 at one atmosphere and 100 ml/min flow rate~ heating rate 5°C/min. Coke was
prepared by pyrolys is of 5% S coal in He at 900 ° C for one hour.
-------
2. KINETICS OF COAL DESULFURIZA TION
A complete kinetic treatment, in terms of a set of elementary
heterogenous and homogenous reactions and the evaluation of the
frequency factors and activation energies of each reaction, is pre-
eluded by the immense complexity of the system.
Nonetheless, a
practical kinetic picture of this complex system that provides a use-
ful framework within which to consider design of a large scale de-
sulfurization process may be obtained. To develop this picture we
consider the evolution rates of hydrogen sulfide (since it is by far
the most abundant sulfur-containing gas) in terms of the four types
of process discussed in the theoretical section, specifically
Aso\..
It
~
13,.11'"
...
1-' 1,. !
( 11)
A,~\;~ .. \-\1 ~'f' ~h"1 ."... ~ ~~ ( 201)
')
A ,. I. ~ R .) 1\.1.,,'" + ~ ".1 (271)
( R
b
I-\~,..., + IJ\ ~\o) B,.,...,.. ~~S (421)
~ "
'-
-58-
-------
Analysis of the hydrogen sulfide evolution curves in hydrogen
(a typical run being shown in Figure 8) on the basis of the irreversible
processes (11) and (201) and by means of the theoretical treatment
described in Section II lead to the conclusion that six separate pro-
cesses are required to give a reasonable theoretical fit to the data,
each process being characterized by a value of ko and E. The ko value
for these studies may be only pseudo first-order but because of the large
[Hz] is indistinguishable from true first-order.
Certainly in the case
of the lowest-temperature process to produce hydrogen sulfide the rate
constant determined does depend on [Hz] , since, as explained, a meas-
urab1e shift occurs when helium is substituted for hydrogen.
The re-
suits obtained from this simplest case of irreversible reaction are
shown as apparent frequency factors and apparent activation energies
in Table VIII.
Since the apparent kinetic parameters were obtained on the basis
of reactions (11) and (201), one may attribute validity to them only if
the reverse reactions of the hydrogen sulfide with the coke are negligible.
That such back-reaction complications are probably important has been
d 1f . . 26. h' h .t
suggested in earlier studies of coal esu unzatlOn, ln w lC 1 was
observed that some of the forward reactions (desulfurization) are accel-
erated by an increase in flow rate of the flush gas. Moreover, investiga-
tions of the inhibiting effect of hydrogen sulfide on desulfurization of
low-temperature chars27, 28 have shown quite conclusively that the
reverse reactions of hydrogen sulfide with coke are important.
_rfJ-
-------
ILl
CI)
z
o
CI. . !OOO
CI)
ILl
cr
cr
ILl
....
ILl
I ~
<:1'0
o cr
I ....
()
ILl
0-
CI)
CI)
CI)
ct
~
~o.)o
~ooo
2000
1000
o
o
200
100
!OO
Figure
8.
2
l~CALCULATED
6
EXPERIMENTAL
\
400
500
600
700
800
900
1000
1100
TEMPERATURE
(0C)
AnC\lysis of the hydrogen sulfide expcrinlcntal results on the basis
of separate first- order irrevers ible reactions.
-------
TABLE VIII: APPARENT KINETIC PARAMETERS FOR HYDROGEN
SULFIDE EVOLUTION IN THE NON-ISOTHERMAL PYROLYSIS OF
5% SULFUR COAL IN HYDROGEN ATMOSPHERE, RUN N1
Reaction No. To ( .. C) E( kca1/mo1e) ko( min-l) % of T ota1
1 225 26 5 x 1010 2
2 380 25 3 x 107 43.5
3 510 27 3 x 106 7.5
4 650 33 5 x 106 8
5 875 43 1 x 107 15
6 1000 69 5 x 10 10 24
-61-
-------
To assess further the possibility of this complication, a series
of non-isothermal experiments in hydrogen were conducted at different
flow rates of hydrogen, ranging from 100 ml/min to 800 ml/min (Runs
Nl, 10, 11, 12).
The hydrogen sulfide evolution curves obtained in
these experiments are shown in Figure 9.
The effect of flow rate of flush
gas on the prominent high-temperature peak is quite dramatic showing a
shift to lower temperatures as the flow rate is increased.
The effe ct
on the prominent low-temperature peak is not nearly so pronounced
(if indeed it is real) and is opposite in direction.
Such shifts with flow
rate are in accord with the occurrence of significant back reactions.
Furthermore, the direction of the shift is predicted by equation (41)
to depend upon whether E-Eb is equal to or greater than zero.
The temperature of the high-temperature maximum is plotted as
a function of the mean residence time 1 in Figure 10. The extrapolation
to 1-=. 0 (infinite flow rate) shows that the position of the high-tempera-
ture evolution-rate maximum would be in the range of 600-6SO°C if no
back reaction were occurring.
InFigure 11 are shown according to
Sections II-2 and II-4 two actual extreme cases, one for no back reaction
and one for a fast back reaction. The change of shape, as well as position
of the maximum, with flow rate shows that neglect of back reaction of
hydrogen sulfide with some components of the solid coke can yield com-
pletel y misleading results.
To confirm directly the occurrence of a back reaction of hydrogen
sulfide with the coke, we prepared a sample of coke from the original
coal by pyrolysis in hydrogen at five atmospheres pressure for 1.8
hours, using a hydrogen flow rate of 800 std ml/min. We then passed
-62-
-------
8
7
- ~-""~-"-":'-~5-~~~L'-<"~' :_"'7";:--
,--.- -. -. ~ TM-~ -
100 800
6
:u
1"1
~
<
1"1
." 5
:u
0
CD
~
!!!
r-
-i
I ~ 4
0'
W
I ~
:u
CD
c
Z
-i 3
foG
-
2
o
100
Figure
200
300
1500 600
TEMPERATURE ( 0 C)
900
700
800
1000
1100
9.
HzS evolution curve for different hydrogen flow rates.
identifying the curve is hydrogen flow r~te in ml/min.
The par<'lmeter
-------
I
'"
~
I
3.0
2.0
~
-
.
.
ft
1.0
800
700
Figure 10.
100 1ft 1 1.1"
400 ..I I.....
800 .1 I....
r Colc.lated U8i...
/ 0' Bock RooctiOfl
800
Unci., E8tlmate
900
1000
TEMPERATURE
( PEAK)
Location of the high temperature peak in the HzS evolution in
hydrogen as a function of mean residence time, 1'" .
-------
..2
1.0
NO BACK REACTION
0.8
~
,.
III
:II o.t
o
n
I
0'
U1
I
0.4
0.2
300
400
500
FAST BACK REACTION~
1<2" » I
~--.
,/ .......
'"
.-- ,
/ ~,
/' \
,/ \
/' \
,/
/
/"
.,/
.......
-..- -
--
-
600
700
TEMPERATURE
(0 c t
800
900
1000
/100
Figure
Cornparison of the HzS evolution curve for a fast back
reaction with the calculated curve for the case of no back
reaction.
u.
-------
a flush gas, comprised of 1000 ppm of hydrogen sulfide in heliUIn, over
the coke in a non-isothermal experiment.
Monitoring the concentration
of hydrogen sulfide in the flush gas as a function of temperature showed
conclusively that hydrogen sulfide reacts very rapidly with the coke
produced and does so over the temperature range of interest in desulf-
urization kinetics.
In Figure 12 is shown a plot of the outlet concen-
trations of hydrogen sulfide as a function of temperature.
The dif-
ference between inlet and outlet hydrogen sulfide concentrations is
proportional to the probability of absorption of hydrogen sulfide in the
bed and we shall as surne that the disappearance is due to back reaction
with coke.
As can be seen from Figure 12 the hydrogen sulfide con-
sumption begins at about 300°C, rises to a peak at 425°, declines
slightly, and then rises continuously until at 775° all of the hydrogen
sulfide in the feed gas is consumed by the bed.
of this experiment are depicted in Table IX.
The integrated results
In the discussion of the theoretical basis of our experimental
methods we showed in equation (48) that for such a situation we may
w rite
A (~ [~~Slo)
,co~ [~Lsl
-
-
ioa~k'T) -
*'a ( J. ,
~.1.J &1 TJ
.
( 48)
Therefore, a log-In plot of the ratio of the inlet and outlet hydrogen
s.ulfide concentrations versus 4 in accord with equation (481) (Figure 13)
. ld . h . ~ iJ
Y1e s a stra1g t l1ne whose slope is - 'Z., 303R. and whose intercept at
1 .
T = 0 1S log kbo": From this plot we obtain the following kinetic para-
meters for the back reaction of hydrogen sulfide with coke:
-66-
-------
HzS
250
0-.
-.]
I
200
1
z
g
!c
c
...
z
...
u
z
o
u
...
~
!c
..J
...
C
50
HELIUM
.
.... ..
. ..
................
.....~....... .
. ......
........................
He
70(1)
~0QC)
.
INLET ..
H S .
CONClNTRATION
. .-..---....
..
.
o
. ....
... ..
.. .
.
.
..
..
.
.
.
....
..
.
.
.
800
1100
100
200
300
400
500 600
TEMPERATURE (0 C)
700
900
1000
Figure 12.
Relative peak heights for hydrogen sulfide and helium sampled from the
effluent of a bed of coke for Run N22. The carrier gas contained 0.103%
hydrogen sulfide in helium.
-------
TABLE IX BACK REACTION OF HYDROGEN SULFIDE WITH COKE
HzS Con- Temp. °C Solid
Run centra - Heating Res idence Time Residence, 1 ( 4)
No. Sample Wt, g tion Rate Reaction Time % Coa
N22 coke 1.130 0.1% 25-1100 0.14 sec. 95. 1 4.2
from in 33° /min 2.4 hr.
Run N 20 helium to 200
70 /min to
1100
Sulfur 0/0
k (5)
Co e
Coke
H zS ( 6)
1.9
4.2
Notes
( 1) 20/40 me sh coke.
(2) quartz reactor.
(3) purge gas at 1 atmosphere pressure
( 4) in coal sample from which coke was prepared.
(5) in coke sample before reaction with HzS.
(6) in coke sample after reaction with HzS.
I
0"
(XI
I
TABLE X
PYRITE PYROL YSIS EXPERIMENTAL DA TA FOR NON -ISOTHERMAL
HYDROGEN RUN
Temp. °C Solid
Run Wt. g Pur ge Heating Residence Time Res idence, Solid
No. Sample %S Gas Rate Reaction Time % Residue
N9 pyrite 0.252 hydrogen 25-1100 0.25sec. 41.6 < 0.1
52.5 5 ° /min 3.8 hr.
Sulfur Recovery, %
SOz
CH3SH Total
HzS CSz C4H4S Recovery
102 < 0.1 < 0.1 102
Notes
( 1)
( 2)
( 3)
100/200 mesh pyrite.
quarts reactor:.
purge gas at 1 atmosphere pressure.
-------
I
0'
..L)
I
log (-In 3)
10 \ Co
0.5
.o.S
- ).0
-).3
-;2.0
0.9
1.0
o
0°
00 0
.00
o
o
1.0
'.2
/.7
/. 5" /. ~
/lJi'r ( .K)
/.1
/.5
/.6
Figure 13.
Arrhenius type plot of the data from the direct measurement of the hydrogen sulfide back
reaction with coke from Run N22.
-------
, 1.1
\tca.' 1-',
E~:
\t~. ~
r . -8 -t
~ -t 10m'''' \ .....
With the kinetic parameters, determined as above, for the back
reaction of hydrogen sulfide with coke, we may use the theoretical
treatments of Sections II- 3 and 11-4 to obtain the kinetic parameters
of the desulfurization (forward reaction) using non-reactive and re-
active flush gases, respectively.
For example, application of the treatment of Section U-4 to
the actual high-temperature hydrogen sulfide evolution peak obtained
in the experiment with a flow rate of 100 ml/min depicted in Figure 9
yields directly the values
E - *..
-
-
~9. i IIc..' 1..(.
n.
't"~o
In this particular experiment 1'": 3 seconds.
-
-
'1
,. 2 t& I 0
. -, -,
1'111. 3 .
From "t and the kinetic
parameters of the back reaction, namely, kbo and Eb, we obtain from
these data the kinetic parameters of this desulfurization reaction as
".=
e I' . -, -,
I.... ~ 10 ...,.. ,
"1\ ftc..I/..J.
E:
-70-
-------
When a sample of pure pyrite was desulfurized in a non-isothermal
experiment using hydrogen as flush gas, we obtain a sharp evolution
peak of hydrogen sulfide at about 5000 and a broad lower pe:::.k at a
higher temperature. The experimental results are shown in Figure
14 and Table X. We conclude that the peak at 5000 corresponds to
reduction of pyrite to sulfide and that the higher temperature peak
corresponds to reaction of the sulfide with hydrogen to form hydrogen
sulfide. Since non-isothermal experiments on coal samples show
consistently a sharp peak at 3800, we conclude that this lowest-
temperature peak in coal desulfurization is due to attack of hydrogen
on organic sulfur.
Thus, we have identified three separate type-
reactions that produce hydrogen sulfide from coal in a hydrogen at-
mosphere.
The kinetic parameters determined for these processes,
as just des cribed for the hydrogen sulfide formation from sulfide,
are shown as "Indirectly Determined" values in Table XI.
Using
these kinetic parameters the "expected" three peaks in a coal
desulfurization non-isothermal experiment are calculated and shown
in Figure 15.
In Se ction II- 5 we pointed out that an alternative way to obtain
the kinetic parameters of the desulfurization reaction (forward reaction)
would be to measure the apparent values (such as given in Table XI) for
various flow rates and extrapolate the apparent values to infinite flow rate
-71-
~
-------
100
78
.
.
MASS SPECTROMETER
.
.
u '50
0 .
o
0 GAS CHROMATOGRAPH
.....
I ..
-J N
....' :II:
117' ~
I .
...
0 .
...
#- 28
.
.
.
.
.
o
100
200
500
400
.
.
.... ....-....
500 600
TEMPERATURE (0 C)
.
.
.el.
700
800
900
1000
1100
-------
TABLE XI. KINETIC PARAMETERS FOR THE DESULFURIZATION REACTIONS
REAC TION NUMBER 1 2 3
IDENTIFICA TION Organic-S {\ n FeSz + Hz FeS + Hz
(Tentative) 0 ales :> HzS ~HzS
). HzS
I
-..J INDIRE C T
N
I DETERMINA TION
E (kca1/rno1e) 36.5 43
ko (rnin -1) 1.9 x 109 1. 9 x 108
DIRE C T
MEASUREMENT
E (kca1/rno1e) 32 38 <;:4
ko( rnin -1) 4.2 x 109 9 . 4 x 109 2.7x1010
-------
~ 0.4
~
~
..
F 0.1
.,.
c
:8
I "
-J 0
\.N n
I
0.2
0.1
0.8
0.1
100
100
O,..nlo
Sui\:
400
500
Figure 15.
100 700
TEMPERATURE (0 C)
800
900
1000
HzS evolution curve calculated by the indirect method.
1100
-------
or zero residence time. In order to attempt this extrapolation we
have studied in Run N23 the evolution curves of hydrogen sulfide in
non-isothermal experiments for a residence of 0.03 seconds. This
residence time is 100 times smaller than that used in the desnlfurization
experiment, des cribed above, that permitted calculations of the kinetic
parameters of the formation of hydrogen sulfide from sulfide.
In this
short residence time experiment, the results of which are shown in
Figure 16, we find the three peaks expected, in accord with the three
kinetic reaction-types identified and shown in Table XI.
The locations
of these peaks are at 400°C, 500°C and 590°C in reasonable agreement
with the expectations shown in Figure 15.
Above 650°C there is a tail
on the hydrogen sulfide evolution curve extending to 800 ° C, which is
apparently due to some back reaction occurring even at this short
residence time.
Analysis of these three peaks by the theory described
in Section II-2, using the assumption that r:::0.03 seconds corresponds
to an extrapolation to 1:.0, leads to the kinetic parameters shown as
"Directly Determined" in Table XI. Considering the extreme rapidity
of the back reaction and the as sumptions involved in both methods of
treatment we consider the agreement as satisfactory.
In Figure 17
the rates of the three desulfurization reactions and the hydrogen sulfide
back reaction are shown as functions of the temperature, using the
"direct" determination.
-74-
-------
:u
m
r
~
-i
<
m
o
o
z
o
11'1
Z
-f
.~
~
o
z
I
-.,J
U1
I
60
50
40
30
:r
<
...
20
10
400
500
o
100
300
1
TEMPERATURE
.
.
.
.
.
.
600
.
.
...
....,
'........
,
800
700
900
(0 C)
Fi:gure 16.
~omp~rison of the theoretical curves for HzS evolution in
hydrogen with no back reaction with the experimental result
from Run NZ3.
'000
IIOC.
-------
10"
Rate Cons taht
10'"
I.. gS (as HzS)m~n-~
\gS ( in bed) .)
101
500
700
1000
1100
800
800 lOa
TEMPERATURE (0 C )
Rates for the desulfurization reactions and the HzS
back reaction as a function of temperature in
hydrogen at one atmospher~.
-76-
Figure 17.
-------
In the short residence time experiment just des cribed, the coke
produced was analyzed for combustible sulfur by combustion in oxygen
at 1l00°C. The condensable gases were trapped at liquid nitrogen
temperature and analyzed for sulfur dioxide using the gas chromato-
graph. No sulfur dioxide was detected when the sensitivity limit
corresponded to 0.01 milligrams of sulfur as sulfur dioxide. This
result indicates that at least 99.9% of the combustible sulfur was removed
by the reaction with hydrogen.
3. FORMS OF SULFUR
The results obtained under conditions of minimal back reaction, or
correction for it,indicate that the original assumptions concerning which
forms of sulfur are involved in the three desulfurization reactions were
somewhat in error.
The total organic sulfur cannot be accounted for by
the gas evolution peak occurring at 400°C; rather it appears that this
peak corresponds to the organic sulfur associated with the volatiles and
that the remainder of the organic sulfur is removed along with the in-
Za
Powell has
organic sulfide by the reaction peaking at 590 ° C.
studied the variation of forms of sulfur in the coke as a function of
carbonization temperature for several different coals.
His results
for a high-volatile bituminous Tennes see coal containing approximately
4% sulfur are given in Figure 18 and show the complete reduction of the
sulfate below 500°C and the reduction of the pyrite below 5Z5°C. There
-77-
-,.t,J
-------
~ I
i
n
i
-4
...
Z
-4
-
I
I .
-..J .
00 ..
I
#-
0
41
I
-
o
P"lt.
1.If.t.
-/
~.... -
----~:
~:P
~/,- -
100
zoo
500
400
Figure 18.
0''''
/
\
\
ZHZI
i-
I
,-,
I" '.... lulflde
...... / --
......- ----
.....~ - -- --
-
/TO,
500
800
TOO
800
'00
1000
TEMPERATURE (0 C)
Forms of sulfur in coke as a function of carbonization
temperature, from Powell,Reference 2a.
-------
is a minimum. in the organic sulfur content at about 370°C.
Above
600°C his results show essentially no variation in the forms of sulfur
with increasing temperature.
From his results and ours it appears
to be pos sible to as sign forms of sulfur to each of our three observed
reactions. The lower temperature reaction, peaking at 400°C,
corresponds to reduction of sulfur associated with the volatile
organic material.
The second peak at 500 ° C, as shown by our measure-
ments on pyrite, corresponds to the reduction of pyrite to sulfide.
Also,
at a temperature below 500°C the sulfate is reduced to sulfide, possibly
accompanied by the evolution of a very small part of the sulfur as sulfur
dioxide.
The process peaking at 590°C corresponds to the reduction
of the sulfide and the organic sulfur associated with the fixed carbon.
At present it is not clear why these two dissimilar reactions apparently
occur as a single process.
The reasons for the persistence of organic
sulfur and inorganic sulfide in the coke even at high reaction temperatures
are clear from our measurements on the kinetics of the back reaction.
. 2a f d
Under the conditions at which Powell's expenments were per orme ,
and under the conditions of most of our experiments, the back reaction
of hydrogen sulfide with the coke is so fast at high temperatures that
the probability of hydrogen sulfide escaping from the bed is extremely
small. The products of these back reactions are apparently both
organic sulfur species and inorganic sulfide.
-79-
-------
v.
CONCLUSIONS
As a direct result of the studies described in this report we have
reached the following conclusions pertinent to development of a practical
coal desulfurization process.
1. The non-isothermal method is an extremely powerful technique for
obtaining the kinetic data neces sary for the rational development of
coal desulfurization processes.
2. The back reaction of hydrogen sulfide with coke is the most important
factor limiting the rate and efficiency of desulfurization.
3.
Practical desulfurization can pos sibly be accomplished during
pyrolysis and without complete gasification by intimately mixing a
suitable sulfur absorbent with the coal in the pyrolyzing reactor.
provided a sulfur absorbent can be found possessing the following
properties:
( a)
The reaction of HzS with the absorbent must be competitive
with the reactions of HzS with coke under the conditions for
which the primary desulfurization reactions proceed at
practical rates.
( b)
Reactions of hydrogen and other gases present in the reactor
with the sulfur-laden absorbent to desorb sulfur compounds
must be slow compared to the absorption reactions.
( c)
The absorbent must be such that the sulfur-free coke can
be physically separated from the absorbent or such that
the coke can be gasified without causing the release of the
sulfur housed by the absorbent.
-80-
-------
4. The rate of the slowest desulfurization reaction is sufficiently
fast at 750°C and one atmosphere of hydrogen that at least 990/0
desulfurization can be accomplished with a reaction time of 0.5
minutes, provided the sulfur absorbent competes effectively with the
coke for the hydrogen sulfide under these conditions. The desulfuriza-
tion reaction rates are increased at higher hydrogen pressures and the
frequency factors for the reactions are probably proportional to hydrogen
pres sure in the range near one atmosphere.
However, precise measure-
ments of the kinetic effect of hydrogen pressure is obscured in the pre-
sent results by the effect of the back reaction. When high-volatile
bituminous coals are used, the hydrogen produced from the coal may
be an adequate source.
5. Below about 0.1 mm the effect of particle size on the reaction kinetics
appears to be small. The small effect of particle size observed in this
work is possibly due to variation in the actual residence time used in
the studies on particle size effects.
Since the effe ct of the back re-
action was not understood at the time these experiments were conducted,
adequate control over the appropriate reaction variables was not
exercised.
6.
Heating rate effects are probably due to the fast back reaction.
At
fast heating rates the velocity of the back reaction becomes very large
before any substantial amount of hydrogen sulfide can es cape from the
bed, thus slowing the overall rate of the desulfurization reactions.
In pyrolysis of coal mixed with a suitable sulfur absorbent the effect
of heating rate should be much smaller.
These conclusions indicate that desulfurization during coal gasifica-
tions prior to combustion for power may be feasible and possibly econom-
ically practical, provided a suitable sulfur absorbent can be found.
- 81-
-------
The recent work of Squires 29 suggests
possess many of the re quired properties.
that calcined dolomites
Substantial additional work
is required to elucidate further the kinetics of the reactions involved in
the desulfurization of coal; to establish the generality with which the
present results apply to a variety of coals; to identify suitable high-
temperature sulfur absorbents; and to determine the kinetics of the
pertinent absorption and regeneration reactions.
Also additional data
on the kinetics of coal gasification reactions will probably be required
to complete the laboratory evaluation and development of a practical
process.
-82-
-------
11.
12.
13.
14.
15.
16.
VI.
REFERENCES
1. (a)
R.R. Lowry, editor, If Chemistry of Coal Utilization", Vo1s. I
and II, John Wiley and Sons, Inc., New York 1945.
(b) R.R. Lowry, editor, , If Chemistry of Coal Utilization I: , Supplement,
John Wiley and Sons, Inc.. New York 1963.
D. W. Van Kreve1en, "Coa11f, Elsevier Publishing Company,
New York, 1961.
2. (a) Powell, Ind. and Eng. Chern. 12, 1069 (1920) .
(b) Powell, Ind. and Eng. Chern. 12, 1077 ( 1920) .
3. Snow, Ind. and Eng. Chern. 24, 903 (1932) .
( c)
4. Foerster and Geissler, Z. Angew Chern.~, 193 (1922) .
5. Sperr, Proc. 2nd Inter. Con£. Bituminous Coal I, 560 (1928) .
6. Ditz and Wildner, Brennstoff - Chern. 2' 149 (1924) .
7. Campbell, Bull. Am. Inst. Mining Eng. 1916,
177.
8. Wibaut and Stoffel, Rec. Trav. Chim. 38, 132 (1919) .
9. Monkhouse and Cobb, Gas J. 156, l34 (1921) .
10. Wibaut, Rec. Trav. Chirn. 38, 159 (1919) .
Wibaut, Brennstoff-Chem. 3, 273 ( 1922) .
Parr and Layng, Mining and Met. 1920, No. 158, Sec. 4.
McCallum, Chern. Eng. .!!' No.1, 27 (1910).
Wibaut and LaBastide, Rec. Trav. Chi"rn. 43, 731 (1924) .
Juntgen, Erdo1 and Koh1e.!,2, 180 ( 1964) .
Van Reek, Juntgen, and Peters, Brennstoff-Chern. 48, No.6, 35 (1967)
Translation by Scientific Research Instruments, Inc-:-:- Baltimore, 1968.
17.
18.
Peters and J\lntgen, Brennstoff - Chernie 46, 175 ( 1965) .
Van Reek, Juntgen, and Peters, Ber. Bunsen. Phys. 2!., 113 (1967).
Van Reek, Juntgen, and Peters, "Theoretische und experirnentelle
Vorstudien zur Kinetik der Koh1enwissenschaftliche Tagung, Munster
1. bis 3, June 1965.
19.
20.
21.
Juntgen and Traenckner, Brennstoff-Chem. 45, 105 (1964) .
Ji'intgen and van Reek, Fue147, 103 (1968) .
-83- ".
-------
22. Hanbaba, J\intgen, and Peters, Ber. Bunsenges. phys. Chern. 1968,
72 (to be published) .
23. Button, Greg, and Winsor, Trans. Faraday Soc. 48, 63 ( 1952) .
24. Pechkovski and Zvedin, C.A. 56, 5460 (1962).
25. Pagurova, "Tables of the Exponential Integral, "Pergarnon Press, New
York, 1961.
26. Mason, Ind. Eng. Chern. 51, 1027 ( 1959) .
27.
Zielke, Curran, Gorin, and Goring, Ind. Eng. Chern. 46, 53 (1954) .
28. Batchelor, Gorin and Zielke, Ind. Eng. Chern. g, 161 (1960) .
29. Squires. A.M., "Fuel Gasification", Advances in Chemistry, Series 69,
Robert F. Gould, editor (American Chemical Society, Washington, D. C.,
1967 ) Cha pte r 14.
-84-
-------
APPENDIX
During the first four months of this project a laboratory for
measurements on the heterogenous reaction kinetics of coal gasifica-
tion and desulfurization was constructed. Photographs o.f this lab-
oratory are presented in Plates I, II, and Ill.
APPARATUS
The experimental apparatus consisted of three main sections;
the combustor-gasification furnace and gas collection traps, the GLC
sampling and gas handling system, and the analytical section which
included the gas chromatograph, plasma spectrograph, and the mass
spectrometer. A block diagram of this apparatus is shown in Figure
AI. Detailed descriptions of each section are shown in Figures A2-A5.
The reactor used in most of these measurements was construc-
ted from a 75 cm length of 20 mm 1. D. quartz tubing with 1.5 mm
walls as supplied by the Thermal American Fused Quartz Company.
This tubing was sealed at both ends with stainless steel O-ring con-
nectors which were adapted for this purpose from NRC-Norton
Company connectors.
The se fittings were epoxied to the quartz tube
and connected the reactor vessel to the flow system.
Stainless steel
water cooled jackets, which also served as plugs to the reactor tube
were used in the early runs. They consisted of a double sleeve en-
capsulation around 1/4" O.D. tubing formed by 1/2" O.D. and 3/4"
O. D. tubing in a concentric arrangement. Overall length of the water
cooled jackets was 4.5". This cooling effect at the inlet of the reactor
-1-
-------
I
N
I
LABORATORY FOR THE STUDY OF COAL GASIFICATION AND
DESULFURIZATION
PLATE ::t
~H-"-..-.5
-------
,
I
\.N
I
PLASMA SPECTROGRAPH FOR COAL-SULFUR ANALYSES
PLATE 1[
PH-86-68-65
-------
,
.;..
I
PLATE 111:
MASS SPECTROMETER FOR NON-ISOTHERMAL KINETICS OF COAL-
SULFUR CHEMISTRY
PH-..-..-85
-------
I
\J1
I
Ma..
Spoctra.otor
...
HI or H. " F.'..co '" ..
.. ... Callectlo. S,.to. .... Vent
~.
8lC
8oIIIpll., Spt-
,
Pla...a
elC
Spoct,..,aplt
"II-
Ho
-.
Figure~A 1. Block diagram of experimental apparatus.
-------
. Flow MeIer
(
0'
I
. H)'cJrogen
To Gos
Monifold
HOI
Zone
Quortz
Flasks
. lIneor
Temperoture
Controller
To Sconning M05$ Spectrometer
Figure A2. Detailed description of coal pyrolysis gas handling system.
Vlol!:r
Bub~l!:r
-------
To
From
H. Supply
GLC
Lln.
I
-.]
, I
.. 4'.
,",L~ rJ.,
~( X
l.J. ~-
, ,
, I
I I
I I
\ I
\ 8LC 18...,.. Loop I
, "
, /
.... ,
- ...
.......... --'
Valv.
V.nt Volv.
Mon..t... 0-15 p.1
(Dlft. Bourdon)
To
Vacuum
S ,.t...
Fr..z.
Tip No.1
Fr..z.
Tip No.2
Figure A3. Detail schematic of GLC sampling and gas handling system.
-------
S.,II
Collutlon
Trope
""'1.. ...
Mondllnl
&yet-
'11. I
I.L.C.
I
CD
I
HI
~
Rlcordlr
To E......t
PI081110
Splctrolroph
Vocuu..
Pu..,
Au.illor,
80.
Suppl,
Figure A4. Detail S>Ghe:rn.a1ic of tandem GLC -Plasma Spectrograph.
Rlcordlr
-------
ION
IEAM
I,
.;0
I
O8.IECT
SLIT (.005. .125)
COLLECTOR SLIT
(.02er. .575)
8A1
INLET
~.
FigureA5. Schematic diagram of the ion optics for the
scanning mass spectrorneter.
ION COLLECTOR
-------
tube occasionally caused the coal particles to agglomerate in the inner
tube of the water cooled jacket during a gravity feed, and both jackets
were found to be unnecessary and were eliminated during the later runs
They were replaced by a 3/4" O. D. to 1/4" O. D. stainless steel reduc-
ing fitting which was compatible with the O-ring connecto:rs.
Two other reactor vessels were also used.
A stainless steel
reactor, 3/4" O.D.and .065" wall-thickness, but similar in all other respects to
the quartz reactor described above, was used in Runs 13, 14, and IS.
This tube was discarded because of the possibility of a reaction with
the HzS at high temperatures.
The reactor shown in Fig. A6 was used
for the high-pressure series, Runs N17 to N20.
The ins ide li~e r wa s
a 76 em length of 17.8 rom O. D. quartz tubing with 1.5 mm thick walls from
Thermal American Fused Quartz Company.
O. D. stainless steel tubing with. 083" walls.
The outside jacket was 111
The end plug and elbow
connector were standard 1" Swagelok stainless steel fittings.
The quartz wool used to support the coal in the reactor was approx-
imately 1 gram of Thermal American Fused Quartz Company quartz
mat No. 550 cut and rolled to form a plug. Matheson Company, pre-
purified hydrogen, 99.95% minimum purity, or high purity helium,
99.995% minimum purity, were used as the carrier gas. Flow rates
were monitored with a Brooks Instrument Division flowmeter Sho-Rate
"150" Model 1355 with tubes in the R-2-15 series.
The furnace was a Lindberg/Hevi-Duty Model No. 54032 with Control
Console Model 59344. It has a maximum temperature range of 1200 DC
and is capable of control to + 1 DC. This furnace is designed for heat
applications requiring close temperature control, uniformity and fast
heat-up and uses silicon carbide rods as heating elements. The inner
-10-
-------
S"loil UnIon a Plu,
Stain I... StMI "ocII.t
l
Quartz Tub. Lln.r
Quartz Wool Plu,
l
Epa., S.al
~
Carrl.r ,.. Inl.t
Willt., Val"..
s...... Elbow
-.... £alt To ColI.ctlon S,.t..
Aad .... I,"t..,..t.,
"
Figure A6. Diagram of reactor vessel used in high pressure
experiments.
-11-
-------
unit which contains the two semicylindrical heating units is enclosed
in an insulating brick and steel case, and air convection between this
unit and an outer steel shell reduce the surface temperature of the fur-
nace to below 100°C at the maximum temperature.
The heating elements
are 12" long and, because of this long heating path, gas flowing through the
reactor is sufficiently heated before it reaches the coal sample in the center
of the heating zone. In the later runs a quartz wool plug was placed on
top of the coal to furnish additional hot contact with the carrier gas.
This furnace incorporates an API contactless controlling pyrometer
placed at the mid-point of the heating elements as shown in Figure A2.
The position of the pointer of the meter reading the furnace thermocouple
voltage is sensed optically, providing a sensitive yet very reliable design.
For the non-isothermal constant-heating-rate experiments the appara-
tus was modified by adding a linear temperature programme r to the fur-
nace control.
This was a laboratory model designed and built at Scientific
Research Instruments Corporation to be compatible with the Lindberg fur-
nace described previously. In designing the programmer to operate
with the furnace and its controller it was desirable that no changes be
made to the Lindberg unit. The linear programmer shown in Figure A 7
was therefore constructed and has proven very satisfactory in actual use.
The unit generates a linearly increasing voltage of opposite polarity
to the thermocouple output voltage. The generated voltage is in series
with the thermocouple output so that furnace temperature then rises linearly
to maintain zero input voltage to the API controller. Actual furnace tem-
perature can be read any time desired either by connecting a recorder
directly across the thermocouple output or by depressing the READ
TEMPERATURE switch and reading temperature on the meter in the
controller.
-12-
-------
~~FO. ~N"'~OlI
I. S K.
lOOK
I I
- ,
w r-- .
I ' ).,
!
IK rr
aoo.n. 151(
ZERO
:).OK Ao,). 3. 3 ~
R~MP 1.(.11< lO.il
SPEED ~OK
-/a v
COMPENJI\\loN
RAMP
STOP START
o
1010<
+,~ V
4.1 K
3.3~
47o.n.
tI. 7 H.
CIRCUIT
COMMON POINT
T \-\ tR M C> CO\.! P L t-
IN F"URNACE:
.
(MICRo- S't./liCH)
READ
TEMP\;f\ATUR~
.
.
Figure A 7. Schematic diagram of the linear temperature programmer.
+
RECO~DE.R
.
-....(.,
L INo8£RGt
PYROMETER
-------
The ramp generator is basically conventional and begins with an
RC time constant composed of a 311-f low-leakage capacitor and a Z x
109 ohm resistor. Voltage across the capacitor is sensed by a ZN4ZZ1
FET transistor, to provide the necessary high input impedance.
The
following three transistors comprise a differential pair and an emitter
follower. Feedback to the capacitor linearizes the ramp over the 10
volt output range.
Ramp speed is adjustable with the RAMP SPEED
control and may be changed during a run if different linear temperature
rise rates are needed.
Maximum ramp voltage is 10 volts -- a resis-
tance divider ( 1. 5K and 10 ohms) injects a portion of this voltage in
series with the thermocouple.
The controller in the Lindberg furnace incorporates thermocouple
burn-out protection by passing a small current, ii' through the thermo-
couple.
In order to prevent the accuracy of the readings on the Lindberg
controller being affected by the added 10 ohms in the thermocouple cir-
cuit, it is necessary tc) compensate by passing an equal and opposite
current, -ii' through the 10 ohm resistor.
This current can be adjusted
by the COMPENSATION control which is set, with the furnace at room
temperature and ramp control in STOP position, so that no change of
meter defection occurs upon operating the READ TEMPERATURE switch.
In most runs, a Hoke Inc., 75 ml stainless steel sampling cylinder,
No. 4HS75, was adapted for use as a tar trap. Either Hoke 75 or 150 m1
high pressure stainless steel sampling cylinders, No. 4HS75 and 6HS150,
respectively, or standard Pyrex glass vapor traps were used as gas
collection traps. The traps were immersed in liquid nitrogen so that
gases which were condensable in liquid nitrogen were collected, while light
-14-
-------
gases such as methc£ne and carbon monoxide and the carrier gas were
not collected and were vented to the atmosphere. In coal runs No. 113
and 115-118 the traps were arranged in parallel; in all other runs, the
traps were connected in series. After Run No. 110, the traps were 1/2
filled with 3 mm diameter Pyrex beads to increase the collection efficiency.
All components of both the coal gasification and collection system,
shown in detail in Figure A2 and the GLC sampling and gas handling
system shown in Figure A3 were stainless steel.
Unless otherwise
specified, connecting lines were 1/4" O. D. with. 020" wall: ~he valves
used were Whitey Research Tool Co., IVS4-316.
Background pressure
in the gas handling system was maintained at approximately 25f by a
Welch 1400B tw~-stage vacuum pump. This sampling system was also
used for the purification and injection into the GLC of calibration standards
as well as the volatile pyrolysis products.
A carrier flow loop accessory on the chromatograph unit was used
with 1/8" O.D.. .020" wall stainless steel tubing to extend the carrier
flow lines prior to injection in the column so that they could be connected
to the GLC Sample System. This provides a means for gas injections of
samples whose pressures ranged from a few torr to several atmospheres.
'-Injection into the GLC was made by switching the 3-way solenoids, see
Figure A3 so that the helium carrier gas was diverted from the bypass
line to the sample loop sweeping out the gas sample and carrying it to
the columns. The solenoids used were Allied Control Co., Inc. Model
RVSV -30384 stainless steel with Viton seats.
Detailed des cr iptions of the analytical apparatus are shown in
Figures A4 and AS. The chromatograph used was a Varian Aerograph
Model 202-1(; dual column unit with thermal conductivity detectors and
-15-
-------
a linear temperature programmer.
Two 10 ft., 1/4" O.D. stainless
steel columns with 20% Triton 305 on acid washed, DMCS, chromosorb
G were used with the chromatograph.
Operating conditions were:
initial GLC column temperature 50.C. programmed after 5 minutes at
a rate of 4° /min to 70°C; detector temperature, 100°C; injector tem-
perature, 85.C; detector current 200 ma; and helium flow rate, SO ml/min.
Matheson Co. high purity helium was used as the carrier gas. Retention
times for various compounds under these operating conditions are given
in Table AI.
The sulfur-containing products of the coal gasifications were calibrated
under these GLC conditions by injecting into the GLC from a known volume,
gas standards at different known pressures.
The volume used was the
3.5 ml GLC sample loop, see Figure A3; pressures were read on a Wallace
and Tiernan pressure gage Model FA-141.
An example of a calibration
curve for HzS, obtained in this manner. is shown in Figure A8.
This cali-
bration was done using Matheson Co. CP grade HzS.
Other gas calibrations
were made with Matheson Co. sulfur dioxide. commercial grade, and methyl
mercaptan. 99.50/0 minimum purity. J. T. Baker Chemical Co. reagent
grade carbon disulfide was used in the CSz calibration; Aldrich Chemical
Co. technical grade thiophene was used in the C4H,J) calibration. Chroma-
tograms were recorded on a 10" Varian Aerograph Model 20 recorder and
peak areas were measured using a K and E Planimeter Model 62-005.
Using 1.0 cmz as the smallest measurable peak and 115.5 cc as the ex-
pansion volume, the GLC sensitivity was 0.013 mgS. Details of this cal-
culation are shown in the following section. Although 11 5.5 cc was the
expansion volume used when the gas was expanded directly from the gas
trap into the GLC sample loop and larger expansion volumes were often
used when large quantities of gaseous product were present, the smaller
expansion volume represents a lower limit on the sensitivity.
\
-16-
-------
TABLE AI. RETENTION TIMES OF VARIOUS COMPOUNDS IN
GLC FOR TRITON 305 COLUMN PROGRAMMED
FROM 50 - 700 C AND HELIUM FLOW RATE, 50 ML /MIN .
Compound Retention Time (min)
Air 2.1
HzS 3.1
CH4 2.1
2.1
CO
2.1
COz
COS 2.5
14.7
SOz
CH3SH 6.0
CSz 8.8
CJf6 23.0
( CzHs) zS 26.5
C4H4S 40.5
-17-
-------
1500
1000
2000
,
~
00
I
.. 1500
E
.
-
C
III
C
C
~ 1000)
c .
III
..
2500
500
o
o
0.5
1.0
Ficrure ASL Chron-uOli:naT'..n'h r..Hh.....H,.....
_\..._-~
1.5
H-2S PRESSURE
1'-- 1---..:1-- ----
2.0
ht.lea)
2.5
5.0
5.5
4.0
-------
Since 0.013 mg was in all cases equal to or less than 1/10% of the total
sulfur, this GLC sensitivity was more than Suffl'cI'ent f .
or measuring
the sulfur balance. The SO... MeSH CS and C H S h t h'
..' , z, 4 4 C roma ograp IC
sensitivities were similar to that for HzS.
For Runs Il to 113 a plasma spectrograph was connected, in tandem,
to the exit of the chromatograph as shown in Figure A4 d 11 .
an a sma porhon
of the effluent was carried into the plasma detector. The detection
system for monitoring the spectral emission for the helium plasma
included a Jarrell Ash Monochrometer Model 82-410 with a Cs-Sb photo-
multiplier, an electrometer and a Bausch and Lomb one channel strip
chart recorder. A Wratten No. 15 filter with a light transmission cut-
off at 5200 A was installed ahead of the monochrometer to reduce scat-
tered light and increase sensitivity.
Qualitative analysis of sulfurous
gases exiting from the GLC was made by monitoring the 5454 A sulfur
atomic line.
Since spectral studies indicated that this particular sulfur
line was superimposed on a carbon molecular band, a relative indication
of sulfur to carbon atom ratio was made by alternately recording the
5454A sulfur line and the part of the carbon molecular band at 5448A.
This method gave two simultaneous peaks corresponding to sulfur and
carbon concentrations.
An example of this type of plasmagraph is
shown in Figure A9.
Various compounds were run under the conditions described above
to check peak shape sensitivity and retention time. The retention time
for HzS is approximately 3.5 minutes in the GLC and delayed by 15
seconds in the corresponding plasma spectrogram. This delay between
the chromatogram and the spectrogram is the mean gas flow time between
tl,te thermal conductivity detector and the take-off needle valve of the plasma
spectrograph. Since this delay is constant, it did not cause any ambiguity.
-19-
-------
- .
I
N
o
I
>-
e
-
-s.
as
J.4
tIO
o
J.4
..
~
I1c
U).
o
s-
1.If8r At.... ,..
c....
.........
.... ,..
-~----~----
,
7
8
,
"0
/I
1'2..
Minutes
''"Figure ~9. Plasmagram of O. 1 psia methyl mercaptan in 3. 5 cm~ sample loop.
.....
.'
-------
The scanning mass spectrometer used in the non. th
-1S0 ermal con-
stant heating rate experiments was designed and b .It t S. ..
U1 a c1enbf1C
Research Instruments Corporation for this project. For the first two
non-isothermal measurements the mass spectrometer was connected
to the reactor flow system using a membrane separator as shown in
Figure AIO. This separator increased the sensitivity of the measure-
ment substantially for those components to which the membrane was
highly permeable, such as HzS and benzene.
It was found, however,
that this enhancement in sensitivity was not necessary and for all of
the later experiments the membrane separator was replaced by a
stainless steel needle valve.
A schematic diagramof the ion optics of this instrument is given in
Figure AS. The ion trajectories follow a 3" radius of curvature through
900 of magnetic deflection. This instrument uses a fixed ion accelerat-
ing voltage and is magnetically scanned by linearly varying the current
to the magnet using a motor driven potentiometer. For all of the experi.
ments conducted the mass range from mass 1 to mass 84 was scanned
every 50 seconds.
The electrometer connected to the ion collector uses a lOll ohm
resistor and gives a full scale output of 10 volts for 10-10 ampere ion
current.
The time constant for the electrometer is approximately 1
millisecond.
In scanning the spectra the output of the electrometer was
connected to a Brush Mark 280 two channel recorder.
The sensitivities
of the individual channels of the recorder could be selected according to
the ion currents being monitored to cover any range from 10 volts full
scale, down to 100 millivolts. Peaks corresponding to 2 millivolts
were clearly recognizable above noise.
-21-
-------
To Mall ~
Spectramet.r
I
N
N
I
Sila.tic Membran..
)
To VacuUftl Pump
\.
Micro Va Iv.
I
(f
--- Sample Flow
..
Figure AIO. Membrane separator for mass spectrometer inlet system.
-------
In the non-isothermal runs the mass Spe t t .
c rome er contInuously
scanned the mass range I to 84 resulting in a quasi to"
- con Inuous record
of the changes in the composition of the pyrolitic gases with time and
temperature. Calibration factors for the sensitivity of the ........a t
U~ s s spec ro-
I?eter to these gases were determined by measuring the spectrometer
response to a known mixture of the component to be measured and helium.
These known mixtures were prepared on the gas handling system. The
calibration factor relative to helium is then given by
r ~ (Px 'V PHH~
J \J>Hel\ PHx -J
where PHe is the partial pressure of helium in the mixture, Px is the
partial pressure of the component to be measured, PHHe is the spectro-
meter response to helium in millivolts of peak height, and PHx is the
spectrometer response to the component to be measured in millivolts of
peak height.
These calibration factors are listed in Table All relative to
helium. and hydrogen.
The concentrati on of a gas in the volatiles present
at any given time in the pyrolysis was determined by multiplying the
appropriate calibration factor for that gas by the ratio of the peak height
of the mass indicated for that gas in TableAII to the mass 2 or mass 4 peak
height.
PROCEDURE
Basically, the experimental procedure was to pyrolyze coal samples
under different experimental conditions and collect and analyze the evolved
gases, coke and tar for per cent sulfur by weight. The experimental
parameters investigated included type of coal, mesh size, carrier gas,
temperature, heating rate, etc. In addition, in the non-isothermal runs, the
-23-
-------
TABLE AII.MASS SPECTROMETER CALIBRATIONS FOR
THE GASES MEASURED IN TInS WORK
MOLECULE MASS NO. CALIBRATION FACTOR
He ( 1) Hz ( 2 )-
He 4 1 1. 21
Hz 2 0.83 1
C~ 16 0.11 0.13
CO 28 0.053 0.064
CZH6 30 0.17 0.21
HzS 34 0.055 0.067
C3Hs 43 O. 14 0.17
COz 44 0.033 0.040
CH3SH 47 0.038 0.046
50z 64 0.041 0.050
C5z 76 0.012 0.015
C 6l!6 78 0.018 0.022
( 1) Relative to He, calibration factor times ratio of peak height of
indicated mass number to mass 4 peak height equals concentration.
(2) Relative to Hz, mass 2 peak.
-24-
-------
gases evolved were continuously analyzed a f " "
s a unchon of lncreasing
temperature by the scanning mass spectrometer d "b d "
escrl e prevlously.
Two different methods of inserting the coal sample into the reactor
vessel were used. In isothermal fast heating rate experiments, the
coal sample was placed in a section of flexible Tygon tubing at the
inlet of the reactor tube, as shown in FigureA2. After the carrier
gas
flow rate was established and the desired temperature had been reached,
the coal was gravity fed into the furnace by manually lifting the Tygon
tubing so that the coal particles dropped into the furnace through the
water cooled inlet.
In the non-isothermal and slow heating rate experi-
ments, the coal was placed in the furnace prior to the run.
The carrier
gas was turned on and sufficient time allowed for the system to be purged
of air before commencing the heating.
In isothermal pyrolysis runs, experiments were timed from the
coal insertion by gravity feed to the time at which the heat was turned off;
in non-isothermal runs, experiments were timed from beginning to end of
the heat application.
In all cases, helium was used during the cooling
off period, and gases were not collected during this time.
The gases evolved during the pyrolysis were swept through the fur-
nace by the carrier gas and out to the collection traps.
In coal Runs Il3
and Il5-Il8 where the traps were arranged in parallel, the gases evolved
during the first 15 minutes were collected in one trap and the gases
evolved during the remainder of the pyrolysis were collected in another
trap.
In Runs NI-N9 the gases were collected in a series of 12 separate
traps at 20 minute intervals by alternately changing traps.
In all other
runs, two traps were connected in series, and either the gases in one
trap transferred to, the other before analysis, or they were analyzed
separatel y and the results combined.
-25-
-------
After the pyrolysis was completed, the collection traps were
vacuum-purged to remove completely the non-condensable gases,
sealed off, and allowed to warm to room temperature. In Runs I1-I18
the gas was then expanded from the collection trap through a gas inlet
on the GLC sampling system into the GLC sample loop shown in Figure A.3.
In all other runs, this method was replaced by one in which the gas traps
were removed from the collection system and attached directly to the
GLC sample line eliminating the need to expand the samples through a
long segment of stainless steel tubing.
Also, in all other runs, where
large enough quantities of vapor had been collected in the traps for com-
pounds to be present as a liquid, the material was transferred by freezing
with liquid nitrogen into a larger volume, usually either a 1 liter or 5
liter bulb.
This reduced the error due to HzS and other vapors being
dissolved in the liquids such as benzene, CSz, etc., which were present
in varying degree. After the gas was expanded into GLC loop, the sample
was injected into the gas chromatograph and analyzed for sulfur.
An example of the calculation to determine the amount of sulfur in
the HzS gaseous product is shown below for Run N3.
Data: measured HzS peak area - 8539 cmz
HzS chromatograph calibration factor to convert peak area to psi
in the loop (inverse slope of line in Figure A8) -1.26 x 10 -3 psia
crnz
expansion volume used - 115. 5 cc
Calculation:
Step 1 - using the HzS calibration factor to convert the measured
peak area to psia in the loop gives
8539 cmZ x 1. 26 x 10 -3 psia
cm;.!:
:: 10.76 psia
-26-
-------
Step 2 - us in&: the calculated psia and the Perfect Gas Law to
calculate the moles of HzS in the loop ( 3.5 cc) gives
PV
RT =n
=10.4 X 10-7 moles HzS
Step 3 -
using the calculated number of moles HzS in the sample
loop and knowing the ratio of the sampled gas to the total
volume of gas gives
115.5 cc
10.4 x 10 -7 moles x 32g/mole x
3.5 cc
=
109.8 mg S
Quantities of sulfur in the other gaseous products were similarly
calculated. Note that by using an as sumed area of 1.0 cmz in the
above calculation, the GLC sensitivity to HzS is found to be 0.013 mg S
as previously mentioned.
The coke was weighed and analyzed for per cent sulfur by standard
ASTM methods.
The reactor tube and the tar trap were rinsed out
several times with acetone to collect the tar.
The acetone was then
evaporated and the tar was weighed and analyzed for per cent sulfur
by the A TSM methods.
The sulfur balance and per cent sulfur recovery could then be
easily determined from the total sulfur found in the gas, coke, and tar.
-27-
-------
AP7,.5-24
1969
Vestal Ma
AUTHOR Kinetic Studies on the Pyr-
01 sis Desulfurizati & . -
TITLE cation of Coals with Emphasls
on the Non-Isothermal Kinetic
J:v1e,tiho RE~~~~ED
LOANED
AP?.5-24
1969
Vestal, Marvin L. et al
Kinetic Studies on the Pyrolysis,
Desulfurization, & Gasification of
Coals with Emphasis on the Non-Iso-
thermal Kinetic Method
------- |