WATER POLLUTION CONTROL RESEARCH SERIES
17020 DDC 06/71
Effect of Porous Structure
on Carbon Activation
U.S. ENVIRONMENTAL PROTECTION AGENCY
-------
WATER POLLUTION CONTROL RESEARCH SERIES
The Water Pollution Control Research Series describes
the results and progress in the control and abatement
of pollution in our Nation's waters. They provide a
central source of information on the research, develop-
ment, and demonstration activities in the Environmental
Protection Agency, through inhouse research and grants
and contracts with Federal, State, and local agencies,
research institutions, and industrial organizations.
Inquiries pertaining to Water Pollution Control Research
Reports should be directed,to the Chief, Publications
Branch, Research Information Division, Research and
Monitoring, Environmental Protection Agency, Washington,
D. C. 20460.
-------
EFFECT OF POROUS STRUCTURE
ON CARBON ACTIVATION
by
University of Colorado
Chemical Engineering Department
Boulder Colorado
for the
ENVIRONMENTAL PROTECTION AGENCY
Project #17020 DDC
June 1971
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 - Price $1.00
-------
EPA Review Notice
This report has been reviewed by the Environmental
Protection Agency and approved for publication.
Approval does not signify that the contents neces-
sarily reflect the views and policies of the
Environmental Protection Agency; nor does mention
of trade names or commercial products constitute
endorsement or recommendation for use.
13.
-------
ABSTRACT
Reaction rates and porous structures of a calcined Wyoming coal
activated by air and by carbon dioxide and a graphite activated
by carbon dioxide were measured. Total, macropore and micropore
volumes, surface area and pore-size distributions were determined
as functions of burnoff.
In 1-mm-diameter calcined coal particles, the overall reaction
rate was chemical rate limited at temperatures as high as 961° C
and rates as high as 106%/hr. In 19-mm-diameter graphite cylinders,
the rate at 1030 C was diffusion limited. In graphite, pore growth
by end-burning of crystal-lites was an important mechanism. In
calcined coal, pore-volume changes occurred due to heating, carbon
burnoff and the opening of closed pores present in the starting
material. Burnoff of carbon occurred essentially entirely in macro-
pores. Up to 127o burnoff, the closed pores that were opened were
micropores ; above 1270 burnoff they were macropores .
In order to develop large micropore volumes and hence large surface
areas by gas activation, a micropore structure must be present in
the starting material.
This report was submitted in fulfillment of Project Number 17020 DDC,
under the partial sponsorship of the Environmental Protection Agency.
111
-------
CONTENTS
SECTION PAGE
I CONCLUSIONS 1
II RECOMMENDATIONS 3
III INTRODUCTION 5
IV MATERIALS AND METHODS 11
V RESULTS 23
VI DISCUSSION 51
VII ACKNOWLEDGEMENTS 71
VHI REFERENCES 73
IX LIST OF PUBLICATIONS 77
X GLOSSARY 79
XI APPENDICES 83
-------
FIGURES pAGE
1 SCHEMATIC DIAGRAM OF REACTION SYSTEM 12
2 SAMPLE HOLDING BASKET 14
3 LOG OF REACTION RATE VERSUS RECIPROCAL OF ABSOLUTE
TEMPERATURE 25
4 RATIO OF SAMPLE-TO-INITIAL MERCURY DENSITY
VERSUS BURNOFF 35
5 HELIUM DENSITY VERSUS BURNOFF 36
6 MICROPORE-VOLUME RATIO VERSUS BURNOFF 38
7 CUMULATIVE MACROPORE VOLUME VERSUS RADIUS 40
8 NORMALIZED CUMULATIVE MACROPORE VOLUME
VERSUS RADIUS 41
9 MACRO PORE-VOLUME DISTRIBUTION VERSUS RADIUS 42
10 CUMULATIVE MICROPORE VOLUME VERSUS RADIUS 43
11 MICROPORE-VOLUME DISTRIBUTION VERSUS RADIUS 45
12 MEASURED AND CALCULATED RATIOS OF TOTAL PORE
VOLUME VERSUS BURNOFF 53
13 MEASURED AND CALCULATED MICROPORE VOLUMES
VERSUS BURNOFF 55
14 MICROPORE SURFACE AREA AS A FUNCTION OF MICRO-
PORE VOLUME 59
15 MEASURED AND CALCULATED MACROPORE VOLUMES
VERSUS BURNOFF 61
16 PREDICTED AND MEASURED NORMALIZED MACROPORE
VOLUMES VERSUS RADIUS 64
17 PREDICTED AND MEASURED, NORMALIZED, MACROPORE-
VOLUME DISTRIBUTIONS VERSUS RADIUS 65
18 SCHEMATIC DIAGRAM OF CARBON PARTICLE 86
VI
-------
TABLES
NO. PAGE
1 PROXIMATE ANALYSIS OF COAL AND CALCINATE 19
2 ULTIMATE ANALYSIS OF COAL AND CALCINATE 20
3 EFFECT OF PARTICLE SIZE AND GAS FLOW RATE ON
REACTION RATE 24
4 RATE-COEFFICIENT PARAMETERS FOR REACTION OF
CALCINED COAL WITH CARBON DIOXIDE 27
5 EFFECTIVENESS FACTORS 28
6 PORE-VOLUME CHANGES DUE TO HEATING 30
7 RESULTS OF ACTIVATION OF CALCINED COAL 32
7a TWO-STEP BURNING RESULTS ON BASIS OF ORIGINAL
UNBURNED CALCINATE 34
8 MICROPORE VOLUMES 46
9 SURFACE AREAS 47
10 RELATIVE INCREASES OF PORE VOLUME IN GRAPHITE
DUE TO ACTIVATION 48
11 CALCULATED AND MEASURED TOTAL MICROPORE VOLUMES
FOR CARBON DIOXIDE ACTIVATION 56
12 MICROPORE VOLUME AND SURFACE AREA 58
13 MEASURED AND CALCULATED MACROPORE VOLUMES
FOR CARBON DIOXIDE ACTIVATION 62
vu
-------
SECTION I
CONCLUSIONS
1. In the activation of 1-mm-diameter particles of a Wyoming coal with
carbon dioxide, the rate of reaction was chemical reaction controlled up to
reaction rates of 106%/hr and temperatures as high as 961° C.
2. The pore structure of the calcined coal was affected by heating in nitrogen
or helium at temperatures as low as 880°C. The effect of heating was
generally to cause total pore volume and macropore (pores greater than
100 A radius) volume to increase and micropore (pores less than 100 A radius)
volume to decrease.
3. In the activation of purified graphite cylinders (19 mm in diameter) in
carbon dioxide at 1030°C, the overall reaction rate was diffusion limited.
Pore growth in graphite occurred by end-burning of graphite crystallites and
by wall-burning of pores. The former predominated as long as crystallites
were available.
4. Gas activation of the calcined coal resulted in increases in the pore volume
and surface area by removal of carbon and by making accessible closed pores
present in the starting material.
5. Essentially all removal of carbon by burning occurred in macropores.
6. At burnoffs up to 12% in air and up to 400°C, the closed pores that were
opened by activation contributed to the micropore volume. At burnoffs of 15%
and greater in carbon dioxide , and at 800°C and above, the closed pores that
were opened contributed to the macropore volume.
7. The surface of porous carbons was not uniformly reactive to oxidizing
gases. This was probably due to a non-uniform distribution of reactive sites.
8. In order to develop large micropore volumes and hence large surface areas
in the gas-activation of coal, it is necessary that the starting material contain
a substantial network of micropores since few new micropores are formed by
burning.
-------
SECTION n
RE COMMENDATIONS
Pore growth in gas-activated carbons can be described but not yet quantitatively
predicted. The surface of carbons appears to be non-uniformly reactive to
oxidizing gases. Further progress in understanding and predicting pore growth
will require investigation of the nature and distribution of reactive sites on the
surface.
Pore growth is influenced by preheating the carbon, by the activating gas
composition, and by the temperature of activation. Preheating tends to
decrease micropore volume. It is recommended that in activation and re-
activation of carbons, preheating in non-reactive gases be minimized.
Micropore-volume increases occur mainly at low burnoffs. It is therefore
recommended that the maximum burnoff in activation should probably be about
15 to 20%.
In the activation of graphite, the end-burning of graphite crystallites to
produce pores of sizes comparable to the crystallites is an important mechanism.
This suggests that activation of carbon structures containing crystallites of a
given size range could be a means of producing an activated carbon with pores
of that size range. It is recommended that studies to determine the feasibility
of such a procedure be conducted.
-------
SECTION HI
INTRODUCTION
Recognizing the potential value of activated carbons for removal of water
pollutants, the present study was initiated for the general purpose of obtaining
information on the effect of the porous structure of carbon on carbon activation.
It was the basic postulate of this investigation that the initial porous structure
of a carbon exerts a profound influence on the course of the activation reaction
and consequently on the pore structure of the activated carbon product. The
term "porous structure" is used here to mean the surface area, pore volume,
and the distribution of pore volume and surface area with respect to pore size.
The aims of this work were: 1) to study the effects of porous structure of
the starting material on the rate of carbon reaction, and 2) to study the
relationships between the porous structure of the starting material and the
porous structure of the resulting activated carbon. The basic raw material
was a calcinate produced from an Elkol coal from Wyoming. The principal
variables were reaction temperature, gas composition and porous structure
of the carbon. The results for porous structure and reaction rates were
interpreted on the basis of additional studies, done as part of this work, on
reaction kinetics and diffusion. It was anticipated that the information
resulting from this work could be useful in commercial activation processes to
suggest means of producing carbons with "tailored" porous structures and
means of reducing the costs of producing activated carbon.
Adsorbent carbons have a major role in wastewater treatment processes to
meet improved water quality standards. Carbon adsorption as a tertiary
treatment process to remove refractory organic substances from domestic
sewage and industrial waters has been demonstrated in several cases (1. 2, 3) .
Recent process developments have shown the applicability of carbon adsorption.
combined with coagulation-filtration, as a potential replacement for traditional
secondary (biological) treatment, as well as accomplishing refractory
organic removal (4) .
The widespread use of activated carbons in water pollutant removal is limited
by its expense (1.5) . Activated carbons are relatively expensive, ranging
in cost from about $0. 07 to over $0. 70 per Ib (6). The cost of carbon-
adsorption wastewater-treatment processes usually is strongly influenced by
the initial and replacement cost of carbon (1,2) . The quantity, and thus cost,
of carbon is dependent upon its capacity (i. e. , the amount of pollutant adsorbed
per unit mass of carbon) which in turn determines the frequency with which the
carbon must be regenerated, and upon the amount of adsorptive capacity lost
during regeneration.
-------
Activated carbon is a class of carbonaceous materials with the property of
adsorbing various chemical compounds from either the gaseous or liquid
state. Activated carbons can be made from virtually any carbon-containing
substance and have been made from many substances of living origin, ranging
from wood to blood and from petroleum coke to coconut shells (7). The most
notable common characteristic of activated carbons is a very high internal
surface area per unit mass, ranging from perhaps 100 to 2, 000 m2/g , but
generally in the neighborhood of 1, 000 m2/g . In order to obtain such high
areas, it is necessary that the carbon be pierced by a network of very fine
openings or pores.
The preparation of an activated carbon from a carbon source generally involves
three main steps: grinding or compacting to give a desired particle size,
carbonization and activation. Carbonization is the heating of the raw material
in an inert atmosphere to drive off volatile substances and increase the pro-
portion of carbon. Changes in both the porous structure and chemical proper-
ties of the material occur during carbonization. Activation is the reaction of
the carbonized char with some reactant, usually an oxidizing gas (steam,
carbon dioxide or oxygen), to remove portions of the material leaving voids
(pores) behind and to create active functional groups on the surface. The tech-
nology of activation is largely proprietary within the industry (7).
The properties of an activated carbon are determined by the raw material from
which it is made and the process by which it is made. The raw materials are
generally relatively inexpensive; e. g. , coals which typically cost of-the-order
of $5 per ton. The cost of activated carbon is then due mainly to the process-
ing required to convert the raw material to activated form.
Coal is an attractive raw material for activated carbon production, due to its
low cost and great abundance. Some work has been done on coal itself as an
adsorbent for wastewater treatment (8), but the use of coal does not seem to
be attractive (9). Some commercial activated carbons are prepared from
coal (6) .
The properties of activated carbon that are of immediate importance in
determining its usefulness include: cost, physical strength, adsorptive
capacity and particle size. The adsorptive capacity is related to the surface
area per unit mass and the interaction of the surface with specific adsorbates;
it is determined by the inherent properties of the carbon and by the degree
of activation. Descriptive properties of carbons most often measured are:
surface area and the adsorptive capacity for particular adsorbents. For
liquid-phase adsorbent carbons, iodine number, molasses number, methylene
blue number and alkyl benzene sulfonate (ABS) number are frequently re-
ported. For any particular application it is necessary to test various carbons
-------
to find an appropriate one because, while the usual tests are useful indicators,
none can completely predict the performance of a carbon for any adsorbate (6).
The amount of surface area per unit mass of carbon, the accessibility of that
surface to adsorbate molecules and the concentration on the surface of sites
capable of attracting and holding specific adsorbates are the keys to the
adsorptive capacity of a carbon. Surface areas are routinely measured,
usually by the classical Brunauer-Emmett-Teller (BET) method, from gas
adsorption data (10). The accessibility of surface and the interaction of the
surface with a particular adsorbate are difficult to determine. Recent studies
have been successful in directly identifying certain functional groups on carbon
surfaces (11), and this type of information promises to enhance the under-
standing of carbon adsorption.
There have been several studies of the influence of porous structure on the
adsorptive properties of carbon and other adsorbents. Most of the work has
been for physical adsorption of gases (12); indeed, gas adsorption is one of
the principle tools for the analysis of porous structure. The equilibrium
amount of gas physically adsorbed on a solid surface is directly related to
the amount of surface area accessible to the gas molecules. The amounts of
certain solutes adsorbed from aqueous solution have been correlated with the
cumulative surface area of pores greater than specific sizes; e. g. , iodine
number has been reported to correlate with area of pores greater than 10A
diameter, and molasses number with area of pores greater than 28A diameter
(13). In other cases, a correlation between the amount of a specific solute
adsorbed and surface area has not been obtained; e. g. , acetic acid and
hydrogen ion (14,17). The lack of correlation may be due to incomplete
knowledge of the porous structure of the adsorbent or due to a specific inter-
action between the solute and functional groups on the carbon surface. It
would seem possible that functional groups are not uniformly distributed on
the surface.
The situation can be expected to be even more difficult when a mixture of
solutes (and possibly colloidal particles) is present, such as in domestic
sewage or many industrial wastes. Not only may a variety of molecular and
particle sizes be present, but species may compete for adsorption sites,
and there may be only a gross measure of species concentrations such as
COD, BOD or TOC . In view of this, it is not surprising that the adsorptive
capacity for sewage constituents does not correlate well with surface area (6).
Clearly, the tailor- making of adsorbents for various adsorbates would require
detailed knowledge of the influence of the porous structure of the adsorbent,
the molecular size of the adsorbate and the mechanism of adsorption.
-------
The rate of uptake of adsorbates is also a distinct function of the porous
structure of the adsorbent and the character of the adsorbate. This is well
established both theoretically and experimentally for gases (15). For liquids,
the theoretical basis is not as well established as for gases, but ordinary
diffusion theory applied to porous solids must take into account the available
diffusion path (and hence the porous structure) even for liquids (16).
Since there is no standard nomenclature for designating ranges of pore sizes,
herein the following definitions will be used: macropores are pores with
radii greater than 0. OlM (lju = 10~ m = 104 A) ; micropores are pores with
radii less than 0. OlM .
The development of porous structure in carbons as a consequence of activation
reactions has been studied by several investigators. Dubinin and coworkers
attempted to correlate the pore volume with total burnoff, but they had only
partial success (18). Cameron and Stacey reported the removal of fine con-
strictions in pore openings in a carbon reacted with steam (19). Walker and
Raats observed uniform surface reaction in electrode carbon rods, and in-
creases in both macropore and micropore volumes (20). These earlier
results indicated that pore development depended upon the character of the
starting carbon, but did not succeed in giving quantitative interpretations of
pore growth.
Recently there have been a number of papers on pore development in various
carbons. Turkdogan et al., studying a metallurgical coke with rather little
micropore area, found growth to occur principally by increase in macropore
volume (21). The surface area of this coke increased with weight loss up to
about 20%, then decreased. They concluded that none of the usual measure-
ments of porous structure gives a direct measure of the effective surface area;
i. e. , the area which participates in oxidation reactions. They noted that in
small carbon particles (less than 0. 5-mm diameter) and at reaction temper-
atures below 1000°C, internal diffusion is not rate-limiting, yet they also
concluded that reaction probably did not occur in micropores due to diffusion
limitations.
Using microscopic examination, Kadlec et al. compared the pores produced
by zinc dichloride activation with those produced by steam activation and con-
cluded that zinc dichloride produced "bottle-shaped" pores, while steam pro-
duced "conical" pores which were larger at the pore mouth due to a higher
activation rate at the pore mouth (22). Lamond et al. found that very small
carbon-black particles, activated by oxygen, resulted principally in the
growth of pores of less than 9 A diameter (23).
Dubinin and Plavnik observed some growth of micropores in an anthracite
coal activated with steam at 900°C (24). They found that as burn-off pro-
ceeded, larger micropores became more predominant. They concluded that
-------
the size of crystallites determined the size of the pores and that the micro-
pores grew larger as burning proceeded.
Chiche and coworkers found, in the activation of two carbonized coals with
steam at 900°C, that the accessible surface area increased with carbon burn-
off up to about 40% and then decreased (25) . The surface area revealed by
low-angle X-ray diffraction changed little, however, indicating that area
present in the initial coke, but inaccessible to nitrogen, was being exposed by
activation. They also found that pore volume increased up to 40 to 50% burn-
off and that the volume increase was mostly in the micropores. They conclud-
ed that the development of large surface areas upon activation required the
existence of a network of fine micropores in the starting material.
Donnett et al. studied the pore structure of a coal partially oxidized by oxygen
at 225-325° C, carbonized at 650° C and then activated with steam at 900° C
(25) . They found that activation increased pore volume, especially micro-
pore volume, and nitrogen surface area. The surface area increased due to
greater accessibility of micropores and growth of micropores.
These several studies give a complex picture of pore development during
activation. In most cases, pore diffusion seems not to have been important,
though this was not often clearly established. In some cases accessible
micropore volume increased, in others it changed little, but in all cases the
surface area measured by gas absorption increased. The increase in surface
area seemed in most cases to be due to burning which made accessible sur-
faces that were already present in the starting material.
The initial phase of the present work was to react carbonized coal with carbon
dioxide at different conditions, to measure the reaction rates and the resulting
porous structure, and to attempt to relate the results to initial porous struc-
ture and reaction conditions. These early results, combined with kinetic and
diffusion data from the literature, led to the tentative conclusion that pore
diffusion was a major factor in pore growth. In order to determine whether
this was the case, it was necessary to have kinetic data for the specific carbon
being studied and to have more information on diffusion in porous carbons.
The second phase of the work was a detailed study of the kinetics of carbon
dioxide gasification of the carbonized coal. This was accompanied by mea-
surement and analysis of diffusion characteristics of porous carbons. This
work led to the conclusion that the reaction with carbon dioxide took place
almost entirely in the macropores of the carbon wherein it was not diffusion
controlled but was chemical-reaction-rate controlled. This work also in-
dicated that diffusion was probably partially rate limiting in the micropores.
A parallel study of pore growth in large cylinders of a highly pure graphite
showed partial diffusion control of the reaction and pore growth due to end-
burning of crystallites.
-------
The next phase of the work was measurement of pore-growth characteristics
of the carbonized coal over a much wider range of reaction conditions. This
work showed that substantial changes in pore volume occurred when the carbon
was heated using typical reaction conditions, except under a nitrogen or helium
atmosphere such that no reaction took place. Although it was known that pore-
volume changes occur due to heating (27), this result was surprising because
the carbon had previously been heated in nitrogen to temperatures above the
reaction temperature. This led to a study of activation at much lower temper-
atures (300-400° C) where volume changes due to heating alone were not
significant. Air was used as the oxidant at the lower temperatures.
The final phase of this work was the analysis of the results in an attempt to
quantitatively describe the observed pore-growth characteristics. It was not
found possible to develop a theory to completely describe the observations.
The theoretical aspects of this work will be presented in appropriate subsequent
sections.
The balance of this report includes a description of materials and methods
used, presentation of experimental observations, and analysis and discussion
of the results.
10
-------
SECTION IV
MATERIALS AND METHODS
Several different types of experiments and measurements were carried out in
this work. These included: heat treatment of calcined coal, reaction of cal-
cined coal over a wide range of temperatures and gas compositions, reaction
of a purified graphite, diffusion measurements, and measurement of the porous
structure of the various carbons used and produced. The experimental and
measurement techniques, the properties of the material used and the ranges of
the experimental conditions are described in this section.
Heat Treatment
The Elkol, Wyoming coal used as the principal raw material in this work had
been calcined by heating to approximately 790 °C before it was received. The
activation reactions conducted in this work were most often carried out at
temperatures above 790°C. It was found, as had been reported earlier (27),
that in order to obtain consistent, reproducible reaction rate data, it was
necessary to use as a starting material in the reaction a carbon that had been
heated in inert gas to a temperature above the subsequent activation temperature.
For this reason, batches of the calcined coal were heated in a muffle furnace
under an atmosphere of flowing nitrogen to a temperature in the range 1100 to
1300°C for a period of from 1 -* to 16 hours. The heating was conducted
for a long enough period so that the carbon reached a constant weight. A 5%
weight loss was observed when the calcined coal was heated at 1100°C for 16
hours.
Activation Syjstem
Carbon samples were reacted with gases in the reaction system shown
schematically in Figure I. This reaction system was designed and constructed
for this study. Slight modifications were made as necessary for specific
purposes.
The reactor consisted of a Mullite tube (G) of 1-inch inside diameter and
33-inches long, heated by a vertical Globar furnace (N). The lower half of
the tube was packed with porcelain chips to smooth the gas flow and heat the
gas to the reactor temperature. At the bottom end of the reactor tube there
were two openings, one for the gas inlet and the other for a thermocouple.
For the reaction of granular, calcined coal, the sample container was a plati-
num basket (R), 3/4 -inch diameter and 3/4-inch high, with a 30-U. S. S. -
mesh platinum screen as the bottom. The basket was suspended from a plati-
num wire which was connected to it through support wires. The basket was
kept 1/4-inch above the tip of the thermocouple (J) . The details of the
11
-------
PROCESS
ELECTRICAL
THERMOCOUPLE
to
He CO
\
L
:
rvi
J
,
^
V2
i r*n
A
\
V3
W
<
I
L_
c
V4
c
u
_
H
_J
R
C
<
1
5
F— •
r
%
\
\
m^f^^mm
C
E
Hhl
F
U
L^
D
[5-
T
[ZI
D
c!
p
r—
:
0
k fc
\
>
1
I
I
1
I
N
— — "
/vwvJ
u M
__-- — R
22
0
J
FIGURE 1. SCHEMATIC DIAGRAM OF REACTION SYSTEM
-------
basket are shown in Figure 2. Graphite cylinders were suspended in the
furnace on a nichrome wire through a center hole. The top and bottom of the
cylinders were sealed with sauereisen ceramic cement. Samples of calcined
coal ranged in weight from 100 to 500 mg. Smaller samples were used at
higher reaction rates where the measured rate showed some dependence on
sample size, apparently due to temperature gradients in the sample.
The temperature in the reactor tube was measured with a platinum vs. plati-
num-rhodium or a chromel vs. alumel thermocouple (J) extending from the
bottom of the tube up to a point where the hot junction was about 1/2 inch
above the surface of the packing. The temperature of the thermocouple was
read on a Honeywell Model R7161B indicator-controller (K), or a Leeds &
Northrup potentiometer (W). The thermocouple was calibrated against the
melting point of reagent sodium chloride, 800. 6° C.
The controller actuated a relay (L) which in turn controlled the power input to
the furnace through a variable transformer (M). The actual sample temper-
ature was measured by an optical pyrometer (T), via a mirror (S), when the
temperature was above 700°C. The pyrometer reading was set with respect
to the calibrated thermocouple. At lower temperatures the thermocouple was
used to indicate the sample temperature, after it was calibrated by comparing
its reading to that of another thermocouple actually inserted in the carbon bed
and held in place by tying it to the support wires of the basket. When the
carbon was heated with air, the sample temperature was greater than that of
the thermocouple by less than 5° C. at 300° C, and by about 50° C at 400° C.
Temperature values were read to + 1° C and varied during a given run by at
most ± 3°C .
The sample-basket suspension wire was attached to one arm of an analytical
balance (0). A Fisher Automatic Recording Balance Accessory (P) contin-
uously balanced the weight and sent a signal to a Leeds and Northrup Millivolt
Recorder (Q) which recorded the weight on chart paper. The balance was also
used to obtain manually the initial, final and several intermediate weights of
the sample. The balance was sensitive to 0.1 mg.
All the gases except air were supplied from commercial pressure cylinders.
Air was obtained from the laboratory compressed air main. The gases were
purified by passage through a silica gel adsorber (A) to remove moisture.
Then all gases except air were passed through a bed of copper wool in a
stainless steel tube (B) in an electric furnace (H). The temperature was
controlled at 500° C by an on-off controller (I) . The copper was used to
remove oxygen from the gas stream. The gas was then cooled to room
temperature by passing it through a coil which was placed in the cold water
bath (C). The flow of gas was controlled by the valve (D) and was measured
by measuring the pressure drop through a calibrated orifice (E) by means of
a mercury manometer (F). When gas mixtures were used, the individual
13
-------
Support Wires,
B&S 20 Ga.
Pt Foil,
B&S 20 Ga.
Pt Screen,
USS 30 Mesh
/ i \
3/-1'
FIGURE 2. SAMPLE HOLDING BASKET
14
-------
gases were metered through calibrated rotameters with the flow controlled by
valves(V) to give the desired composition. The total flow was then measured
by the orifice flow meter. A total gas flow of 100 cc/sec at STP was used. A
mercury trap was inserted in the gas line downstream of the flow meter to pre-
vent mercury from entering the hot furnace. The gases exiting the reactor
tube were removed by a ventilating system (U)
The procedure for a typical activation run will be described.
First, the ventilation blower (U) and the controller (I) were turned on. A
small flow of hydrogen was sent through the stainless steel tube (B) to reduce
any copper oxide formed during the previous run. After the reduction of
copper oxide was completed, the hydrogen flow was stopped and the system was
flushed with nitrogen.
The controller was set at the required furnace temperature, and the power to
the furnace was turned on. When the furnace reached the set temperature, the
temperature of the chromel-alumel thermocouple junction was measured by the
optical pyrometer. In this way the optical pyrometer was calibrated in the
vicinity of the desired temperature.
When the desired reaction temperature had been reached, the empty basket
was suspended in the reactor tube from the analytical balance and was weighed.
A nitrogen flow was started. The carbon sample was placed in the reactor
basket. The basket was then suspended in the reactor tube, and the sample
was heated at reaction temperature to constant weight. After this heating step,
the nitrogen flow was stopped, and the sample and basket were weighed. The
nitrogen flow was again started, and the controller was adjusted to the desired
value. When the temperature of the sample reached the set value and was
maintained constant, the flow of nitrogen was stopped and immediately the flow
of reactant gases through the system was begun. The total gas flow in all the
experiments was 100 cc/sec at STP. This flow rate was found to be well
below the fluidization velocity of the carbon particles but essentially eliminated
any exterior-film mass-transfer resistance.
The weight of the sample was recorded continuously. Since the total gas flow,
gas composition and temperature were constant throughout the experiment, the
weight loss recorded was only due to the loss of carbon by the chemical re-
action.
When the sample weight had decreased by the desired amount, the flow of
reactant gases was stopped, and the system was flushed with nitrogen. The
power supply to the furnace was then shut off, and the sample was cooled to
about 200° C in nitrogen. The nitrogen flow then was stopped, and the sample
and basket were weighed.
15
-------
The rate of reaction was calculated by the slope of the recorder trace and from
the weight change as determined by initial, periodic and final weighings. The
rates determined in these two ways were found to agree closely. The slope
rates were used in the data analysis.
Diffusion Apparatus
An apparatus was constructed and used to measure apparent diffusivities in
porous materials. No measurements were made on the carbons used in this
work. Typical commercial activated carbons were used to determine the in-
fluence of porous structure on diffusion and the applicability of diffusion
theories. The apparatus is described in detail in the thesis of Allen (28) .
Property Measurements
The determination of the porous structure of carbons prepared from starting
materials of differing structure under different reaction conditions was the
main purpose of this work. The measurement of the structure of porous
solids has been and continues to be the subject of intensive research. An
excellent review of the current status of the subject has been presented recent-
ly by Dullien (29) .
The measurement techniques used in this work for the determination of porous
structure were "helium density", "mercury density", nitrogen-adsorption
isotherms, and mercury penetration. Each method and the interpretation and
analysis of the resulting information will be discussed.
The term "helium density" as used herein is defined as the weight of a solid
sample divided by the volume of helium displaced by £hat sample. The helium-
gas molecule has a diameter of approximately 2.30 A (30). Helium, though
very inert, has been reported to be slightly adsorbed on carbons, but the
error in helium density resulting from adsorption at room temperature is
probably not significant (31) .
Helium densities were determined in a Numinco-Orr Model MIC-101-018
nitrogen-adsorption apparatus. A weighed sample of carbon was degassed by
heating in a sample tube at 200-250°C and 0. 0003-mm Hg pressure for
about 24 hours. The sample was then exposed to helium at room temperature.
The apparatus, method and calculations are described in the manufacturer's
manual (32).
The term "mercury density" as was used herein is defined as the weight of
a solid sample divided by the volume of mercury displaced by that sample. The
mercury density was determined in the low pressur chamber of an Aminco
Model 5-7108 mercury porosimeter. The apparatus, method and calculations
16
-------
are described in the manufacturer's manual (33). A weighed sample was de-
gassed in the vacuum chamber by holding the sample at room temperature for
10 to 15 minutes at a pressure of 0. 005-mm Hg. The mercury density was
calculated as the sample weight divided by the volume of mercury displaced at
an absolute pressure of 1. 7 psia, which corresponds to a pore radius of
about 51 IJL (based on the Laplace equation and values discussed below). Thus,
the mercury density as used herein is the mass of sample divided by the
volume of the sample, including the volume of all pores with radii less than 51(i.
i
Gas adsorption, most commonly nitrogen, is the usual method for examining
the microporous structure of materials. Adsorption at low relative pressures
is used to determine surface area by the classical BET technique (34).
Nitrogen-adsorption measurements were conducted in a Numinco-Orr Model
MIC-101-018 apparatus at liquid-nitrogen temperature. The apparatus and
methods are described in the manufactuerer's manual (32). The BET surface
area is found to depend somewhat on the type of gas molecule due to different
molecular sizes (29) .
As the pressure is increased in a nitrogen-isotherm experiment, the phenome-
non of capillary condensation begins to occur in very small pores, and further
increases in pressure cause it to occur in still larger pores. Thus, nitrogen
isotherms at high relative pressure may be interpreted to give cumulative pore
volume as a function of pore size. To do so requires several assumptions.
First, a pore model (i.e. , a shape) must be assumed. The usual practice
is to assume cylindrically shaped pores, even though microscopic examination
usually indicates irregularily shaped pores (29) . When this is assumed, the
Kelvin equation
, , -2av cos 3
Inp/p0 = —T^F-
may be applied (29) . The properties a, v and cos (3 are taken as the
bulk properties of liquid nitrogen, and the Kelvin equation becomes
rR = -10/4n(p/po) .
The calculations must be corrected for the fact that condensation occurs within
the pores that already have gas adsorbed on their surface, thus reducing the
radius. This is done by the "t-method", using averaged values of t from
several authors (35). The actual pore radius, r, is then the Kelvin radius,
r , plus the appropriate t . The pore volume corresponding to a calculated
radius interval is taken as the total amount of nitrogen desorbed, less the un-
condensed amount desorbed, divided by the density of liquid nitrogen.
Nitrogen-adsorption isotherms in porous carbons typically exhibit hysteresis;
17
-------
i. e. , the adsorption and desorption branches do not coincide. This was found
with'the carbons used in this study and, as is the usual practice (29), the de-
sorption isotherm was used in the calculations. Finally, in very small pores,
that is those less than about 16 A radius, it cannot reasonably be assumed that
a liquid phase is formed since the dimensions of the pore are comparable to the
dimensions of the nitrogen molecule; thus, volume calculations become very
tenuous for calculated radii of less than about 16 A . The nitrogen isotherms
are useful for calculations up to pore radii of 150 to 200 A (36) .
The preceding discussion emphasizes the fact that pore volumes obtained from
nitrogen-adsorption data are not absolute quantities but are subject to limitations
inherent in the several assumptions made in the calculations. Micropore-vol-
ume distributions must be considered, then, relative values. As such they are
useful, however, especially for making comparisons of pore volumes between
similar materials, such as the carbons studied in this work.
Mercury penetration into porous solids under pressure is useful for determining
macropore characteristics. Progressively higher pressures are required to
force mercury into pores of smaller and smaller radius.
An Aminco Model 5-7108 mercury porisimeter was used for these measure-
ments. This instrument is capable of a maximum pressure of 15, 000 psia,
corresponding to a radius of about 0. 006|J. (60 A) . In the later stages of this
work, a Wallace and Tiernan gauge was substituted for the high-pressure gauge
supplied by the manufacturer to allow better accuracy.
It is necessary to make several assumptions in order to convert data for mer-
cury penetration versus pressure into pore volume as a function of pore size.
The important assumptions are: a pore model (i.e. , a shape) for which
straight cylindrical pores are usually assumed; and that the properties of mer-
cury in the pores are the same as the bulk properties.
For cylindrical pores, the radius and pressure are related by the Laplace
equation (37)
-2 a cos 9
r =
P
The properties of mercury used were cr = 475 dynes/cm and 9 = 130° (38).
Substituting these values into the previous equation and converting units gives
= 88.6
P
where the radius, r, is in microns and the pressure, p, is in psia.
The mercury-porisimeter method attributes all the pore volume penetrated
18
-------
between two pressures to pores with radii in the range calculated from those two
pressures by the preceding equation. If the material contains "ink bottle"
shaped pores (that is, volumes of a certain size accessible only through open-
ings of smaller radius), all the corresponding volume will be attributed to the
smaller radius. Dullien has shown recently that this problem can lead, at
least in the case of a porous solid produced by compacting a powder, to a mean
radius (as determined by mercury porisimetry) that is nearly a factor often
lower than that determined by a new microscopic technique (29). This points
out that mercury-penetration results yield relative, rather than absolute, pore-
volume distributions. Again, however, such distributions are useful for com-
paring similar materials such as the carbons studied in this investigation.
The usual interpretations were assumed in this work to use nitrogen-adsorption
data to calculate surface area and cumulative micropore volumes at various
radii (32), and to use mercury-penetration data to calculate cumulative macro-
pore volumes (33). A digital computer program, in FORTRAN IV, was
written to calculate the micropore-volume distributions from the raw nitrogen-
adsorption data.
Materials
The raw material used for most of the activation studies was the calcinate of
a sub-bituminous coal supplied by the FMC Corp. from the Elkol Mine,
Kemmerer, Wyoming. The calcinate was produced from the as-mined coal
by grinding, charring by rapid heating to about 480°C and calcining by
rapid heating to about 790° C. Two different batches of calcinate were used
during the course of this work, but no significant differences were found in
their properties or behavior.
The compositions of the as-mined coal and the calcinate, as given by FMC
Corp. , are given in Table 1.
Table 1 - PROXIMATE ANALYSIS OF COAL AND CALCINATE
Coal Calcinate
Moisture, % 14.1 0.0
Volatile, % 35.6 4.5
Fixed Carbon, % 46.7 89.2
Ash. % 3.6 6.3
Total, % 100. 0 100. 0
19
-------
The ultimate analysis, also supplied by FMC Corp. , is given in Table 2.
Table 2 - ULTIMATE ANALYSIS OF COAL AND CALCINATE pRY BASIS)
C H N S O Ash Total
Coal, wt% 70.8 5.06 1.09 0.80 18.24 4.01 100.0
Calcinate, wt % 89.6 1.27 1.22 0.43 1.66 5.82 100.0
The calcinate, as shipped, was screened and the less-than-16 but greater-than-
20 USS-mesh fraction (nominal particle size 1. 0 mm) was used throughout
this work, except in a few cases for which different mesh sizes are denoted.
Batches of this fraction of calcinate were heat-treated in nitrogen within a
temperature range of 1100-1300°C before use for most of the activation studies.
This will be discussed more fully when the results are presented. By means of
heat treatment and partial activation, starting materials for further activation
were prepared which had different initial porous structures.
A portion of this work involved activation of a graphite. The graphite used was
obtained from the Pure Carbon Company. It was their DS-12 grade with a
reported porosity of 20 volume percent. Pure Carbon Company reported that
the graphite contained a maximum ash content of 0. 001 percent. The graphite
was in the form of cylinders, 19 mm in diameter, 25. 4 mm in length and with
a 6-mm hole through the center.
Diffusion properties were measured on two commercial activated carbons. One
was a North American Carbon, Inc., G-352 activated carbon with a porosity of
0.598, in pellets 4.67-mm diameter by 8.1-mmlong. The second was a
Chemetron Chemical Corp., G-32E activated carbon with a porosity of 0. 508,
in pellets 4. 45-mm diameter and 6. 51-mm long. More detailed structural
data have been reported (28).
Commercial bottled gases were used in this investigation. Carbon dioxide.
nitrogen, helium, and hydrogen were supplied by the Natural Gas Company,
Denver, Colorado. Carbon monoxide was supplied by the Matheson Company,
Inc. The compositions of these gases, as stated by the suppliers, were as
follows:
N : 99. 98 % N0, 0. 02 % O_, Dew point - 70 °F
A & &
He: 99. 998% He, 0. 002 % O
£1
H : 99. 998% EL, 0. 0004% O0, 0. 0004% N Dew point - 67 °F
^ A A &
CO2: 99. 63 % CO2, 0.05%O2i trace combustibles, Dew point - 88°F
CO: 99.5% minimum CO
20
-------
The gases were purified as described earlier. Air was taken from the labora-
tory compressed air supply and was passed through a cotton filter to remove
oil, then through a silica gel bed to remove moisture.
Calculation of Properties
Throughout this report all properties that are based on a unit mass, of carbon
will be presented on the basis of a unit mass of starting material. Starting
material is defined as the carbon to be reacted, with zero weight fraction re-
moved by reacting gas, but after any weight change due to loss of volatiles.
Properties measured per g of sample after weight loss due to reaction were
converted to the corresponding values for one g of starting material by mul-
tiplying these properties by the fraction of the original weight remaining after
reaction. For example, suppose 10% of the weight of starting material was
removed by reaction (1 0% burnoff) , then one g of starting material would
yield 0. 9 g of reacted sample. If the pore volume of the reacted sample was,
say, 0. 5 cc/g , then the pore volume was 0. 5 X 0. 9 = 0. 45 cc/g of starting
material. Putting all the values on the basis of starting material makes it
possible to compare directly the properties of materials which have been re-
acted to different extents of burnoff.
The total pore volume, V , in a sample was determined, in cc/g, from the
mercury and helium densities. The mercury density is the mass of carbon,
divided by the volume of carbon plus the volume of all pores with radii less
than Sip, . The reciprocal of the mercury density is thus the volume of carbon,
plus pores, per unit mass. The reciprocal of the helium density is the vol-
ume of carbon not penetrated by helium, per unit mass. The reciprocal of
the mercury density, minus the reciprocal of the helium density, is thus the
volume penetrated by helium but not by mercury. Hence.
v --L.-L . (1)
P PTT PTT
Hg He
This then gives the volume of all pores ranging from those accessible to, and
penetrable by, helium up to those of 51^ radius.
The porosity, E , is closely related to the total pore volume. The porosity
is the volume of pores, divided by the volume of carbon plus the volume of
pores. It is thus given by
PHe//pHg
= 1 ~ pHg/pHe '
21
(2)
-------
Porosity was not converted to a starting material basis because it is dimension-
less.
Mercury-penetration data were used to calculate macropore volume and macro-
pore-volume distribution. Macropores are herein defined as pores with radii
greater than 0. Olp, (100 A). The dividing line between macropores and micro-
pores is somewhat arbitrary, but this value was found convenient since the
carbons which were studied exhibited bimodal pore distributions with large
volumes in pores with radii greater than 0. Olp, , virtually no volume between
0. Olp, and 0. 005|J. radius, and then pore volume again in pores less than
about 0. 005|j, (50 A) radius. The macropore volume was thus taken as the
volume penetrated by mercury at a pressure corresponding to an 0. 01 u. pore
radius; i. e. , a pressure of about 8860 psia.
Cumulative macropore volumes as a function of radius were determined from
mercury penetration at various pressures, using calculations based on the
Laplace equation as discussed earlier. That is, for any given pressure and
corresponding radius, the cumulative volume of pores with radii greater than
that radius is equal to the total volume of mercury that has penetrated the
sample under that pressure. Results for cumulative volume versus radius were
differentiated graphically or numerically to give volume-distribution curves.
Total micropore volume was calculated as the total pore volume minus the
macropore volume. Cumulative volumes of micropores having radii greater
than a given radius were calculated from the nitrogen-desorption isotherms via
the Kelvin equation as described earlier. The cumulative volume data were
graphically or numerically differentiated to give volume distribution by radius.
Surface area was determined from nitrogen-adsorption data by the usual BET
method (34) .
22
-------
SECTION V
RESULTS
The results of the experiments will be presented in this section. The original,
raw data may be found in the several theses prepared as portions of this work
(28, 35, 39, 40, 41. 42, 43) .
Kinetics and Diffusion
Rate data obtained from the reaction of calcined coal with carbon dioxide were
analyzed to determine the appropriate rate expression and rate coefficients.
The effect of temperature on rate was investigated to establish the rate-control-
ling mechanism of the reaction.
Several interesting features of the carbon-carbon dioxide reaction should be
noted. It was observed with the calcined coal, as with other carbons (48),
that during the initial stages of reaction, the rate increased slowly with time
up to 5 to 10% burnoff. This phenomenon is usually attributed to the exposure
of reactive sites resulting from the removal of carbon (49). The reaction rate
then reached a constant value that did not change as the weight of carbon de-
creased and the surface area of the carbon increased. Because of this, re-
action rates are expressed as the percent decrease in initial weight per hour
during the constant rate period. This is, again, a common observation and is
attributed to the reaction occurring at discrete, active sites which are not
changed by the reaction (49). At high degrees of burnoff, usually those greater
than 50% of the initial weight, the reaction rate may decrease (35).
In the calcined coal used in this work, it was found that nearly all of the
observed weight-loss of carbon occurred in the macropores, even though the
vast majority of the surface area was in the micropores (this will be demon-
strated and discussed later). This again confirms that the reaction rate is not
directly related to total surface area. It further indicates that the controlling
mechanism of the observed reaction rate is essentially entirely that associated
with the macropores.
Reaction rates of calcined coal with carbon dioxide were measured over a
temperature range from 780 to 961° C in gas atmospheres containing from
30 to 100 mol% carbon dioxide, from 0 to 20 mol% carbon monoxide, and
from 0 to 70 mol% inert diluent (either nitrogen or helium) at 620 to
630-mm Hg total pressure. The measured rates ranged from 1. 4 to 106%
per hour.
One observation was that heat-treating the calcinate had a very significant
effect on the rate of reaction. For example calcinate, heated for 1. 5 hrs in
23
-------
nitrogen at 1300°C reacted at 31. 4%/hr with carbon dioxide at 875 °C;
whereas calcinate, heated 16 hrs at 1100°C, reacted at 14. 5%/hr with carbon
dioxide at 875° C (42) . This effect was not studied in detail. However, as
will be shown later pore-volume changes also occurred upon heat treatment.
It seems likely that the form and reactivity of the carbon was changed by heat-
ing as has been reported (27); also pore volume and surface-area changes
occurred which may have affected reactivity. The rate data plotted and dis-
cussed here are for calcined coal heated at IIOO'C for 16 hrs in nitrogen.
Data for other heat-treatment conditions were analyzed, but gave different rate
coefficients.
If the log of the reaction rate is plotted against the reciprocal of absolute tem-
perature, as is illustrated in Figure 3, it is found that below some tempera-
ture (about 880° C in this case) a straight line is obtained, and above that
temperature there is curvature. Such a result has been observed earlier, and
the curvature has frequently been interpreted as being due to the onset of intra-
particle diffusion control of the reaction rate (50). It has also been pointed
out that this inference is weak in the case of a complex reaction such as carbon
gasification (50). Thus, further investigations were conducted to determine
whether diffusion was rate limiting.
Table 3 presents the observed rates of the carbon-carbon dioxide reaction with
different gas-flow rates and different particle sizes. The average particle
size was varied from 1.0 to 0. 62 mm and the Reynolds number was changed
from 32 to 64.
Table 3 - EFFECT OF PARTICLE SIZE
AND GAS FLOW RATE ON REACTION RATE
imperature,
°C
920
920
920
961
961
961
961
Particle Size,
USS mesh
-16+20
-28 +30
-16+20
-16 +20
-20 +28
-20 +28
-28 +30
Gas Flow,
cc/sec, STP
100
100
50
100
100
50
100
Rate,
%/hr
47.5
48.3
47.1
100.
104.
106.
104.
24
-------
1.7
1.5
1.3
1.1
a
0.9
•p
cd
OS
c
o
o
nj
0.7
-------
From these results, it could be seen that the rate was not affected by particle
size or the gas flow rate. The rates at one temperature agree within experi-
mental error, thus indicating that the overall reaction rate was not even par-
tially controlled by pore diffusion or by external gas diffusion.
As a further test of this result, the rate constants for temperatures below
880°C were calculated using the rate expression,
ki pco2
Rate = ——; -: (3)
1+k2PCO+k3PC09
Li
as will be discussed below. Then the rates at the higher temperatures were
calculated from equation 3 using the rate constants obtained from the lower
temperature data. The average, absolute, percent difference of 7. 2% be-
tween the calculated and observed rates at the higher temperatures was less
than the average, absolute, percent difference of 10.1% between the calcu-
lated and observed rates for the lower temperature data from which the con-
stants had been calculated. Thus the reaction rate expression fitted the high
temperature data as well as the low temperature data with the same rate con-
stants, indicating that the same reaction mechanism holds over the entire
temperature range.
From these results it was concluded that the overall reaction rate was con-
trolled by the rate of the chemical reaction over the entire temperature range
investigated and that the curvature in the plot of log of reaction rate versus
reciprocal of absolute temperature was due to the form of the rate equation,
rather than the onset of the diffusion control.
Several rate equations have been proposed for the carbon-carbon dioxide re-
action (42). Eight different rate expressions , including two empirical power
models, were fitted to all the reaction rate-temperature data to find which
would fit best and to evaluate the parameters in the models (42). When this
was done, the rate coefficients, k., were taken to be of the form,
k. = A. (exp (- AE./RT)) (4)
so that data for all temperatures were included. For example, A is the
pre-exponential factor and AE^ the activation energy for rate coefficient k]_.
The calculations were done using a digital computer program based on a non-
linear least-square algorithm (44, 45) which minimized the normal sum-of-
squares. The result of this was that equation 3 fitted the data as well or
better than any other model and with fewer adjustable parameters than equally
well-fitting models. Equation 3 fitted the data with an average absolute
error of 7. 2%, compared to an estimated experimental error of 8. 5%. Thus,
26
-------
equation 3, the generally accepted rate expression for the carbon-carbon
dioxide reaction (51), was found to fit the present data satisfactorily (42).
The fit for various temperatures and gas compositions is shown in Figure 3.
The pre-exponential factors and the activation energies obtained for the rate
coefficients of equation 3 are shown in Table 4.
Table 4 - RATE-COEFFICIENT PARAMETERS FOR
REACTION OF CALCINED COAL WITH CARBON DIOXIDE
CALCINATE HEATED AT 1100° C FOR 16 HRS IN NITROGEN
Quantity Value
Temperature, °C 852 to 961
p atm 0.83 to 0.25
C°2 •
p , atm 0. 00 to 0.17
(-'O
A %/hr-atm 4.9 X 1Q10
E kcal/g mol 48.5
A atm"1 2.5 x io~14
Li
E kcal/g mol -78.5
£
A atm'1 2.3 x 10~24
O
E kcal/g mol -124.0
O
Thus, it was concluded that for the calcined coal used, the observed overall
reaction rate was controlled by the rate of chemical reaction, and not by
diffusion, at temperatures as high as 961° C and rates as high as 106%/hr.
It was further concluded that the rate could be represented by equation 3 with
the rate constants given as a function of temperature by equation 4. The
actual values of the rate constants were found to depend upon the heat treat-
ment of the calcinate prior to reaction (42) .
The foregoing results show that the observed rate of reaction is chemical
reaction limited. However, since the reaction was found to occur almost
entirely in the macropores, the rate-limiting mechanism in the micropores
was not clearly established. That is, the micropores, even with their far
greater surface area than the macropores, seem to react very slowly. Is
27
-------
this because diffusion strongly inhibits reaction in the micropores, or is it
that the micropore surface is inherently less reactive? This question cannot
be answered unambiguously, but the following analysis gives an indication that
diffusion is partially limiting in the micropores.
The reaction-kinetic results developed above, plus information on the diffusion
characteristics of activated carbons (28), may be used to estimate effective-
ness factors for various pore sizes. The effectiveness factor is the ratio of
the reaction rate, considering diffusion, to that without diffusion and is thus
indicative of the extent of diffusion control (46) .
Measurements of unsteady-state diffusion coefficients on two activated carbons,
as well as other porous solids, showed that the effective diffusivity in the solids
always lay between the extremes of Knudsen diffusion and bulk diffusion, and
they were rather well predicted by the theoretical model of Otani et al. (47),
provided adsorption did not occur (28). When adsorption did occur, as with
carbon dioxide and carbon monoxide on carbon, the effective diffusivities
tended to be substantially higher than those predicted by the model (28). In
any event, the predominant diffusion mechanisms appear to be Knudsen dif-
fusion in the micropores and ordinary diffusion in the macropores.
Effectiveness factors for macropores were estimated by the method of Roberts
and Satterfield (46), using the kinetic data from above and using the bulk
diffusivity for macropores and the Knudsen diffusivity for the micropores. The
estimated effectiveness factors are summarized in Table 5.
Table 5 - EFFECTIVENESS FACTORS
t
Temperature, Effectiveness Factor
°C Micropore
Macropore 2Q o ^ ^
852
872
920
961
1.0
1.0
1.0
0.99
0.9
0.8
0.75
0. 75
0.8
0.7
0.6
0.6
The macropore-effectiveness factors support the previous conclusion, that is
that diffusion is not rate limiting up to 961° C; although they suggest that, at
that temperature, diffusion may be just beginning to influence the rate. The
micropore-effectiveness factors, on the other hand, suggest that diffusion is
at least partially rate limiting in the micropores over the temperature range
852 to 961° C and that it may be rather strongly rate limiting in pores with
radii less than 10 A . This partial diffusion limitation probably does not
28
-------
fully explain the relatively slow growth of micropores. This will be further
discussed later.
In the reaction of graphite cylinders of 0.75 inches diameter with carbon dioxide
at 1030° C, which resulted in reaction rates of about 5%/hr, diffusion effects
were noted. The small apparent activation energy of the reaction,
66 kcal/g mol, plus less increase in pore volume as one moved radially in-
ward from the outside surface to the center of the cylinder, established that
diffusion was limiting the internal rate of reaction (40). Some growth of
small pores was observed in the graphite, nonetheless. The graphite, however,
contained essentially no micropores, so it is likely that diffusion rate was not
affected by pore size.
The very great differences in reaction rates between the calcined coal and the
purified graphite should be noted. Whereas the calcined coal reacted with
carbon dioxide at a rate of about 100 %/hr at 960° C, the graphite reacted at
5 %/hr at 1030°C. Such vast differences in reactivity between graphite and
the more amorphous and impure forms of carbon are commonly observed (51).
The reasons are probably that the carbon in the graphitic crystal structure is
more stable and less reactive than amorphous carbon and, perhaps even more
importantly, that many mineral impurities catalyze the oxidation of carbon
(51). The catalytic effect of many metals and metal compounds is well recog-
nized (52).
Effects of Heat Treatment
The calcined coal, as obtained, was screened and then heated in batches in
nitrogen. This was necessary in order to obtain consistent, reproducible
reaction-rate data (27,35). As this study progressed, it was found that
heating in nitrogen affected the rate of reaction when the carbon was subse-
quently reacted with oxidizing gases and that heating in nitrogen also affected
the porous structure of the carbon. Some comparative rates after different
heat treatments were already mentioned. These phenomena were not studied
in detail. However, some results on changes in pore structure due to heat-
ing will be presented here.
Experiments were conducted to determine the effects of heating in either
nitrogen or helium on the pore volumes of the carbon. The results are
presented in Table 6 . The first row in Table 6 shows the properties of the
raw calcinate as received. The last three rows show the properties of the
raw calcinate after heating in nitrogen at temperatures up to 400°C. The
results show that essentially no change in pore structure occurred due to
heating in nitrogen up to 4001.
29
-------
Table 6 - PORE-VOLUME CHANGES DUE TO HEATING
Gas Time of
Starting Temperature, Atmos- Heating, Weight
Material
Raw Cal-
cinate,
unheated
Raw Cal-
cinate
Raw Cal-
cinate
Raw Cal-
cinate
Calcinate
Heated to
1100°C
Calcinate
Heated to
1200°C
Calcinate
Heated to
1200°C
Raw Calci-
nate
Raw Calci -
nate
Raw Calci-
nate
phere Hr
1100 Nitrogen 16
1200 Nitrogen 16
1300 Nitrogen 1.5
880 Helium 1.5
900
900
400
350
300
Helium
10
Nitrogen 10
5.
5.
0. 0
0.0
0.0
Nitrogen 1 to 2 4.
Nitrogen ItolO 3.6
Nitrogen Ito24 3.
Pore Volume, cc/g
of starting material
Macro- Micro -
Change,% Total pore pore
423 .278 .145
.455
473 .431 . 042
580 .470 .110
.565
.593 .560 .033
.458 .376 .082
425 .278 .147
.424 .278 .146
,411 .271 .140
This was true even though there was some weight loss due to devolitalization.
This result was further substantiated by the fact that the macropore cumulative
volumes were essentially unchanged (43).
On the other hand, heating the raw calcinate in nitrogen to temperatures of
1100 to 1300° C did change the pore structure. The total pore volume in-
creased, the macropore volume increased and the micropore volume decreased.
It is interesting to note that all the measurable changes were in mercury
30
-------
density and mercury penetration; the helium density remained constant at
1. 64 to 1. 66 g/cc in all these cases (43). The constant helium density
indicates that closed pores inaccessible to helium were not being opened, nor
were pores originally open being closed and made inaccessible to helium.
Apparently the loss of micropore volume was due either to expansion of micro-
pores to become macropores or to complete collapse of micropores. It is
also possible that the heating changed the surface properties so as to change
the wetting characteristics of mercury, but there is no other evidence for this.
It is not clear why the sample heated to 1200°C showed less micropore vol-
ume than that at 13001, although the heating time of the former was consider-
ably longer.
Further experiments were done in which calcinate that had been heated in
nitrogen to 1100 or 1200°C was reheated in nitrogen or helium at temper-
atures of 880 or 900° C. It was supposed that no changes would occur upon
reheating since the material had already been heated to a higher temper-
ature. It was found, however, as shown in Table 6, that rather major
changes in pore volume occurred upon reheating. This was especially true
when the gaseous atmosphere was helium rather than nitrogen. It should be
noted that in these cases there was no detectable weight loss and no change in
helium density. It should also be noted that the rate of heating in the reheat-
ing process was much greater than that of the original heating. Again, the
reasons for these changes are not clear, but the fact that they occur at
temperatures as low as 880° C could make the interpretation of pore-growth
experiments at higher temperatures more difficult since it must be recog-
nized that pore-volume changes might occur which are not a direct result of
reaction. Apparently either thermal shock or carbon mobility at higher
temperatures caused the observed volume changes due to heating (27).
A significant result of these observations is that heating in inert atmospheres
tends to change the porous structure and in particular tends to reduce both
the amount of micropore volume and the micropore fraction of the total pore
volume. This is an undesirable change if the objective is to make a high-
surface-area activated carbon. This may explain why it is stated that chars
heated above 700 to 800° C do not produce active carbons (7).
Effects of Activation
Calcined coal samples were activated at some 34 different sets of conditions.
Table 7 contains a summary of the results obtained, except for the cumu-
lative macropore and micropore volumes and the volume distributions which
will be presented below. For the last two reaction conditions in Table 7
the starting material had already been partially reacted and the pore vol-
umes and surface areas presented are based upon one g of that starting
material. These same results are presented in Table 7a, but converted to
the basis of one g of original unburned calcinate.
31
-------
Table 7 - RESULTS OF ACTIVATION OF CALCINED COAL
REACTION
CONDITIONS
Air, 300°C
Air, 350°C
Air, 400°C
PERCENT
BURNOFF
ol
2
5
7
8
9
10
12
o1
1.4
4.4
5
6.4
8
8.4
10
12
o1
1
4
5
6
8
10
12
V
g/cc
.986
.970
.949
.921
.927
.898
.901
.868
.976
.968
.937
.953
.906
.921
.870
.889
.845
.975
.964
.925
.941
.890
.878
.875
.829
PHe>
g/cc
1.661
1.664
1.669
1.742
1.790
1.799
1.830
1.848
1.662
1.667
1.678
1.677
1.770
1.825
1.833
1.861
1.875
1.663
1.669
1.680
1.680
1.763
1.847
1.883
1.899
Porosity,
e
.406
.418
.431
.471
.478
.500
.508
.530
.414
.420
.441
.432
.489
.493
.526
.524
.550
.415
.423
.449
.440
.495
.525
.536
.564
PORE VOLUME
cc/g of starting material
Total
.411
.423
.431
.476
.478
.509
.507
.537
.424
.428
.452
.431
.504
.494
.555
.532
.571
.426
.435
.467
.441
.524
.551
.548
.600
Macropore
.277
.291
.290
.325
.294
.328
.301
.315
.280
.306
.322
.289
.342
.302
.348
.319
.340
.284
.319
.327
.293
.350
.335
.325
.361
Micropore
.134
.132
.141
.151
.185
.181
.206
.222
.144
.122
.132
.143
.162
.192
.207
.215
.232
.142
.116
.140
.148
.174
.216
.223
.239
SURFACE AREA,
m /g of start-
ing material
co
to
-------
Table 7 (cont.):
REACTION
CONDITIONS
807o carbon
dioxide + 207o
Helium, 872°C
807o carbon
dioxide +20%
carbon mon-
oxide, 888°C
Carbon diox-
ide, 800QC
Carbon diox-
ide, 900°C
Carbon diox-
ide, 900°C
Carbon diox-
ide, 900°C
PERCENT
BURNOFF
o2
15
30
15
30
03
15
30
o3
15
30
o4
15
30
5
0
15
30
V
g/cc
.84
.69
.62
.84
.74
.62
.69
.63
.52
.62
.53
.47
PHe'
g/cc
1.64
1.79
1.93
1.64
1.83
1.89
1.79
1.84
2.00
1.93
1.87
1.97
Porosity,
e
.49
.61
.68
.49
.59
.67
.61
.66
.74
.68
.72
.76
PORE VOLUME ,
cc/g starting material
Total
.58
.75
.77
.58
.68
.76
.89
.88
.99
1.10
1.15
1.13
Macropore
.445
.637
.725
.655
.730
.47
.66
.70
.47
.58
.65
.78
.79
.91
1.00
1.06
1,03
Micropore
.11
.09
.07
.11
.10
.11
.11
.09
.08
.10
.09
.10
SURFACE AREA,
m^/g starting
material
28
47.
48.
45.
46.
38.
46.
43.
38.
52.
59.
54.
70.
84.
61.
91.
95.
to
CO
1. Starting material was raw, calcined coal after heating to constant weight at reaction temperature,
2. Starting material was calcined coal after heating in nitrogen for 16 hrs at 1100°C.
3. Starting material was calcined coal after heating in nitrogen for 1^ hrs at 1300°C.
4. Starting material was calcined coal ,heated \\ hrs at 1300°C, then reacted to 157o burnoff with car-
bon dioxide at 800°C.
5. Starting material was calcined coal^eated 1% hrs at 1300°C, then reacted to 307=, burnoff with
carbon dioxide at 800°C.
-------
Table 7 a - TWO-STEP BURNING RESULTS ON
BASIS OF ORIGINAL UNBURNED CALCINATE
6
ON BASIS OF UNBURNED CARBON
REACTION PERCENT
CONDITIONS BURNOFF
Unburned
Calcinate
Carbon diox-
ide; 15% burn-
PORE VOLUME, cc/g of unburned
calcinate
Total
Macro
Micro
j
0
27. 8
off at 800 °C, | 40.5
remainder at
900°C
Carbon diox-
ide; 30% burn-
off at 80Q°C,
remainder at
40.5
51.
!
.58
. 75
. 84
.81
.79
.47
.67
.77
.74
.72
.11
. 08
. 07
. 07
.07
SURFACE
AREA, nrVg
of unburned
calcinate
38.
60.
71.
64.
67.
900 °C
6. Original starting material was calcined coal heated in
nitrogen for 1| hrs at 1300 °C .
It can be seen in Table 7 that as the burnoff of carbon increased, the mercury
density of the carbon decreased. It has been shown (see appendix 1 for develop-
ment of equations) that if all the burning occurs on the interior surface of the
carbon particles, the ratio of sample-to-initial mercury density, p /p ,
Hg Hgo
should be a linear function of the fraction of carbon remaining, 1 -f (where
f is the fraction burnoff), with a slope of -1. The data are plotted in Figure4
and show that indeed the results follow the theoretical line for a wide range of
gas compositions, burnoffs and temperatures. There is some scatter but the
trend is quite clear, indicating that the removal of carbon during activation
was substantially all from the interior of the particles, creating new void
volume.
The helium density showed a distinctly different trend, as shown in Figure 5.
Up to 4 to 5% burnoff, the helium density did not change; then between about
5 and 12% burnoff, the helium density rose sharply. At burnoffs higher than
12%, the helium density increased slowly. There is a rather sharp decrease
in helium density between 12 and 15% burnoff. Samples up through 12% burn-
off were prepared by reacting in air at 400° C, or below; while those at
34
-------
1.01
0.9
o
60
EC
Q-
^0.
Pi
4-J
•r-l
W
C
0)
Q
o
0.7
0.6
0.5
O Air activated
• Carbon dioxide activated
I
I
1.0
0.9
1.
0.8 0.7
- fraction burnoff
0.6
0.5
FIGURE 4. RATIO OF SAMPLE-TO-INITIAL MERCURY DENSITY
VERSUS BURNOFF
35
-------
2.0
O Air activated
• Carbon dioxide activated
50.
FIGURE 5. HELIUM DENSITY VERSUS BURNOFF
36
-------
burnoffs of 15% and greater were reacted with carbon dioxide at 800 or
900°C. Thus, the sharp drop in helium density is probably due to the
different burning conditions rather than an effect of burnoff alone. Nonethe-
less, the trend seems consistent. Between 0 and 12% burnoff, there is a
distinct correlation between helium density and micropore volume as may be
seen by comparing Figure 6 with Figure 5. Above 12% burnoff. this re-
lationship does not appear to hold. This correlation will be considered in the
discussion section.
The helium molecule has a diameter of approximately 2.3 A. Thus helium
will penetrate all pores with openings greater than about 2.3 A wide, and the
only spaces helium will not penetrate are the solid itself and "closed" pores
(i. e. , pores with openings smaller than the helium molecule). The only
ways in which the helium density could be increased, then, are: 1) a
change in the solid density due to mobility of the solid, recrystallization, or
due to selective gasification of fractions with densities lower than that of the
remaining solid; or 2) for the solid initially to contain pores impenetrable
by helium and for some of these pores to become "opened" (i.e. , penetrable
by helium).
The first point cannot fully explain the observed helium-density increases.
Changes in the solid density seem to be very unlikely because, when the cal-
cinate was heated in an inert gas to typical reaction temperatures, the
helium density did not change (even when there was a loss of weight). Fur-
ther, in order to explain the shape of the helium density vs. burnoff curve,
Figure 5. it would be necessary that the solid volatilized up to 5% burnoff
have a density of about 1. 66 g/cc, that between 5 and 12% burnoff it have a
density considerably less than 1.0 g/cc, and that above 12% burnoff it have
a density of 1.8 to 2.0 g/cc. There is no basis for postulating such a vari-
ation in solid density. It is concluded, therefore, that the observed in-
creases in helium density were a result of the opening of closed pores. This
conclusion is consistent with the earlier conclusion that the observed increase
in reaction rate at low burnoffs was a result of the exposure of active surface,
presumably by opening closed pores.
Total and macropore volumes increased with increased burnoff. Micropore
volume increased up to 12% burnoff, but at higher burnoffs, it tended to be
about constant or to decrease. The BET surface area showed substantial
increases up to 15% burnoff. Above 15% burnoff, surface area increased
little.
The results in Table 7 show that the macropore volume of the starting
materials ranged from two to ten times greater than the micropore volume.
The results also show that at burnoffs of 15% and greater, virtually all of
the change in pore volume was in the macropores. At burnoffs below 12%,
37
-------
1.8
1.6
I
CO
•H
O
4-1
o
a;
a-
o
"1.0
s
M-l
o
o
0.8
0.6
I
0.0
5.
Burnoff, %
10.
FIGURE 6. MICROPORE-VOLUME RATIO VERSUS BURNOFF
38
-------
the changes in volumes of the macropore and micropores were comparable.
This suggests that burning, at least at higher burnoffs, occurs primarily in
macropores. This will be more clearly demonstrated in the discussion
section.
Typical cumulative-macropore-volume results are shown in Figure 7.
These results show that most of the macropore volume lies in the range with
radii between 0.1 and 10jU and that, as burnoff proceeds, the cumulative
volume curves shift to higher radii. In Figure 8, the data of Figure 7 are
replotted as normalized cumulative macropore volume, defined as
(V - V )/(V - V ) , where V is the cumulative macropore volume
at a given radius, V is the total cumulative macropore volume and V
is the cumulative volume at a radius where the volume curves tend to flatten
out at large radius. The normalized curves are somewhat easier to compare
and are useful for analysis, as will be shown in the discussion section.
The distribution of volume by radius interval is more apparent when macro-
pore-volume distributions are plotted against radius as shown in Figure 9.
The volume distribution, d V /dlnr, is obtained by differentiating the
cumulative volume data with respect to the log of the pore radius, or a nor-
malized distribution may be obtained by differentiating the normalized data.
The differentiation may be performed numerically, graphically, or by
differentiating an analytical expression for the cumulative volume, as will be
presented later. The results shown in Figure 9 are from cumulative volume
data differentiated graphically. It is interesting to note that the volume
distribution curves appear to have the shape of a normal (gaussian) distribu-
tion; that is, normal in the value of the log of the radius. This observation
will be made use of in the discussion section. It is clear that as burnoff
proceeds, the maximum in the volume distribution shifts to higher radii.
Cumulative-volume data may be used to estimate the surface area of the
macropores based on an assumed pore shape. For cylindrical pores, the
surface area, dS., included in a narrow range of pore radii is given by
dS = 2dV /r where dV. is the pore volume in the radius range and r.
i i i i i
is the mean radius over that range. Summing the dS.'s (starting at
large radius) gives the cumulative surface area in pores larger than any
given radius. This has been done for the macropores and micropores,
and the results are presented later in Table 9.
Some cumulative micropore volume results
-------
30 7o burnoff
15 % burnoff
Calcined coal
reacted with carbon
dioxide at 800°C
Starting material
0.01
0.1 1.0
Pore Radius, M-, log scale
FIGURE 7. CUMULATIVE MACROPORE VOLUME VERSUS RADIUS
100,
-------
1.0
t>
a
01
S-i
O
a,
o
•H
4-1
tti
i— (
§
0)
N
•H
30 % burnoff
,0.8
Calcined coal
reacted with carbon
dioxide at 800°C
0.6
Starting material
.4
0.2
0.0
I
I
.01
0.1 1.0
Pore Radius, M1, log scale
10.
FIGURE 8. NORMALIZED CUMULATIVE MACROPORE VOLUME VERSUS RADIUS
-------
.60
to
.50
J..40
c
o
•H
4-J
c .30
C
O
•H
4-J
5 .20
J-l
JJ
CD
•H
O
I
3 .10
.00
15 % burnoff
Calcined coal
reacted with carbon
dioxide at 900°C
30 7o burnoff
Starting material
0.01
FIGURE 9.
0.1 1.0
Pore Radius, M-, log scale
MACR0PORE-VOLUME DISTRIBUTION VERSUS RADIUS
10.
-------
.024
to
O
O
.020
.016
t .012
o
a
o
o
•H
s
£ .008
•H
•P
Oj
.004
.00 L
Calcined coal
reacted with carbon
dioxide at 800°C
Starting material
10. 100.
Pore Radius, A, log scale
1000.
FIGURE 10. CUMULATIVE MICROPORE VOLUME VERSUS RADIUS
-------
interpreting nitrogen-adsorption results for radii less than about 16 A, as
discussed earlier. Micropore-volume-distribution curves may be obtained
from cumulative volume data. These were calculated by numerical differen-
tiation of the cumulative volume data by a digital computer program.
Examples of micropo re-volume-distribution results are shown in Figure 11.
It is to be noted that in all cases, the volume distribution was increasing at
the point where calculations were terminated (Usually 7 K radius), indicating
large volumes of very fine micropores in the activated coal.
The presence of very fine micropores in the coal can be shown in other ways.
The volumes of pores with radii less than 100 A, but large enough to be
penetrated by nitrogen, were calculated from the nitrogen-adsorption data
(41) and are shown in Table 8. For comparison, the volumes of micropores
determined previously from mercury and helium-density measurements and
the volume of pores of greater than 10 A radius from cumulative volume
measurements are also given in Table 8, when available.
Comparison of the micropore volumes as determined from nitrogen adsorp-
tion with those from mercury and helium densities shows the latter to be much
greater than the former. This is apparently due to a large pore volume
accessible to helium, but not nitrogen, though it may also be due to the
assumption that nitrogen in the very fine pores has the same density as bulk-
liquid nitrogen. Also, the total micropore volume determined by nitrogen
adsorption is two to five times greater than the volume of pores in the 10 to
100 A radius range as determined from nitrogen adsorption. These results
are indicative that a substantial portion of the micropore volume consists of
very fine pores with radii less than 10 A .
The results show another interesting point; that is, that while the total
micropore volume tended to decrease slightly with increased burnoff, the
micropore volume determined by nitrogen adsorption nearly always increased
with increased burnoff. Finally, the results indicate that greater increases
in micropore volume, as indicated by nitrogen adsorption, took place with
the addition of helium as compared to carbon monoxide, and at 900° C as
compared to 800° C in pure carbon dioxide. These results are not fully
understood.
For liquid-phase-adsorption carbons,the intermediate-size micropores
(i.e., those greater than about 10 A radius) are probably far more impor-
tant than are the fine micropores included in the volume determined by
helium penetration.
Surface-area results are also of interest; these are shown in Table 9.
44
-------
.06
fl
•a
.05
.04
o
c
c
o
•r-l
CD
•H
p
1
.02
.0]
.0
Calcined coal
reacted with carbon
dioxide at 800°C
' i \ Starting material
30% burnoff
FIGURE 11.
10.
Pore Radius, A, log scale
MICROPORE-VOLUME DISTRIBUTION VERSUS RADIUS
100.
-------
Table 8 - MICROPORE VOLUMES
MICROPORE VOLUME, cc/g of starting material
SAMPLE
PREPARATION
Carbon Diox-
ide, 800° C
Carbon Diox-
ide, 900° C
Carbon Diox-
ide, 15% burn-
off at 800° C,
remainder at
900 °C
Carbon Diox-
ide, 30% burn-
off at 800° C,
remainder at
900°C
80% carbon di-
oxide, 20%
carbon monox-
ide, 888° C
80% carbon di-
oxide, 20%
1 PERCENT
BURNOFF
o1
15
30
15
30
27.8
40.5
40.5
51
2
0
15
30
15
30
TOTAL,
Mercury &
Helium
.11
.09
.07
.10
.11
.08
.07
.07
.07
-
-
-
_
-
Nitrogen
; Adsorption
1 .023
.028
.021
.026
.033
.035
.046
.037
.038
.027 i
.025
.023
.032
.028
i
i °
> 10 A radius
.010
.008
.004
i
1
.008
.012
.014
.023
.014
.015
.008
.007
.006
.012
.010
helium. 872" C
1. Starting material was calcined coal heated
in nitrogen for l£ hrs at 1300°C.
2. Starting material was calcined coal heated
in nitrogen for 16 hrs at 1100° C.
46
-------
Table 9 - SURFACE AREAS
SURFACE AREA, m7g of starting material
SAMPLE PRE-
PARATION
100% Helium,
880°C .
80% CO 20%
Li
CO, 872°C
80% CO2, 20%
He, 888° C
CO 800° C
Li
CO2, 900° C
PERCENT
BURNOFFJ BET
O1 : 28.
0 33.
15 ! 45.
30 46.
15 47.
30 48.
O2 38.
15 46.
30 43.
15 52.
30 60.
Macropore,
cumulative
5.4
3.4
3.0
3.0
3.0
4.6
2.7
1.9
2.6
2.2
Micropore,
cumulative
7.4
6.3
6.8
12.7 '
9.8
7.2
8.7 i
4.3 j
i
9.0 j
15.1
Total, cumu-
lative
12.8
—
9.7
9.8
15.7
12.8
11.8
11.4
6.2
11.6
17.3
1. Starting material was calcined coal heated in nitrogen
for 16 hrs at 1100°C.
2. Starting material was calcined coal heated in nitrogen
for 1| hrs at 1300° C.
3. Cumulative micropore area calculated for pores larger
than 10 A radius and less than 100 K radius.
The results in Table 9 show very large differences between the BET surface
area and the cumulative surface area of pores with radii between 10 and 100 A.
The differences here are greater than those usually observed, but such
differences are not unusual when pores with radii less than 14 A are present
(12, 53). The differences are apparently due to the presence of the small
micropores, plus also probably the assumption of cylindrical pores.
The results for micropore-volume changes upon activation give a picture
which is difficult to analyze. There is relatively little increase in total
micropore volume; indeed , there is a tendency for total micropore volume to
47
-------
decrease at burnoffs higher than 15%. There are, on the other hand, in-
creases in the volume of micropores accessible to nitrogen and increases in
BET surface area. In contrast, there was substantial growth in helium-
penetrable-micropo re volume in calcined coal activated with air at up to
400° C and burnoffs up to 12%. Micropo re growth will be considered further
in the discussion section.
Graphite Activation
The reaction of graphite with carbon dioxide gave interesting results about
pore growth which have been published (54).
The results are summarized in Table 10. As was pointed out earlier,
diffusion was partially rate limiting in the large graphite cylinders. The data
presented in Table 10 are the ratio of the pore volume in a certain pore radius
increment of a reacted sample to the pore volume in that same increment in
the unburned sample, considering only the outer radial one-third of the
cylinder (as obtained by grinding off the inner two-thirds). Similar results
were obtained in the middle and inner thirds, but the development of pore
volume was somewhat slower as the center of the cylinder was approached,
due to diffusion limitations.
Table 10 - RELATIVE INCREASES OF PORE VOLUME
IN GRAPHITE DUE TO ACTIVATION
RATIO OF PORE VOLUME IN REACTED
SAMPLE TO INITIAL PORE VOLUME IN
PORE RADIUS INTERVALS SHOWN
% BURNOFF 0 10 20 30
RADIUS INTERVAL, A
70-200 1123
200-500 1322
500-1,000 1789
1,000-3,000 1 8 10 10
3,000-6,000 1344
6,000-10,000 1122
10,000-20,000 1222
20,000-50,000 1 1.2 1.2 1.7
> 50,000 1226
48
-------
The graphite burning results show that during the0first 20% burnoff, the
most rapid pore growth was in the 500 to 3, 000 A pore-radius range.
Above about 20% burnoff, little or no additional growth occurred in that size
range, but rather the larger pores, especially those of radii greater than
20,000 A (2^) , grew relatively more. This is especially noteworthy when
it is recognized that in the unburned graphite, more than half of the total
pore volume was in the 20,000 to 50,000 A radius range. Similar observa-
tions (i.e. , the relatively more rapid growth of smaller pores) were made
earlier by Walker et al. (48) and by Dubinin et al. (24, 55) - in the former
case for sizes similar to those in the present study, and in the later case for
much smaller pores, ca. 6.8 A.
These observations are explained as follows (54). Graphitized carbon con-
O
tains large crystallites, typically of the order of 1000 A thick in the direction
normal to the basal plane (56). Graphite crystallites are known to be much
more reactive in the direction parallel to the basal plane than in the direction
normal to the basal plane (57). Thus graphite crystallites would tend to burn
parallel to the basal plane, producing pores with dimensions comparable to
the thickness of the original crystal lite. This is apparent in the pore volume
increases in the 500 to 3000 A range in Table 10. This type of burning would
occur preferentially as long as crystallites of the proper size were available
and had faces normal to the basal plane exposed to the reacting gas. Growth
of these pores would terminate abruptly when the supply of such crystallites
was exhausted. Again, rapid termination of growth in the 500 to 3000 A
pore-size range occurred. Only when the crystallites were gone did burning
of carbon from the surface of other pores become the predominant mechanism.
This is shown by the results in Table 10 where large-pore growth predomi-
nates between 20 and 30% burnoff. Indeed up to 20% burnoff, only^about
20% of the total pore-volume increase was in the 20, 000 to 50, 000 A pore-
size range; while from 20 to 30% burnoff, about 70% of the total burnoff
occurred in that size range.
Thus pore growth in graphitized carbon can be rather well explained. The
mechanism suggests a way of producing pores of a desired size in an
activated carbon: that is, start with a material containing a high concen-
tration of graphite crystallites of the size of the desired pores.
49
-------
SECTION VI
DISCUSSION
The results presented in section V indicate that in the reaction of a calcined
coal with air or carbon dioxide, carbon removal occurs primarily in the
macropores. The reaction rate is chemical rate controlled, not diffusion
controlled, in the macropores. There is micropore growth under some con-
ditions. Micropore growth is associated with changes in helium density and
is, at least partially, diffusion limited.
In graphite, pore growth was shown to occur largely by end-burning of graph-
ite crystallites. Graphitic crystallites in the calcined coal are probably about
6 A in thickness (21). If these reacted by end-burning, they would lead to
micropores.
In this section, the changes in pore volume will be examined and compared with
pore burning models in an effort to describe the pore-volume changes which
occur upon activation. The results of this study will be compared with other
recent investigations of activation and reactivation.
Total Pore Volume
The total pore volume in calcined coal was found to increase with increased
burnoff of carbon. Increased burnoff was accompanied by increases in
helium density which were ascribed to the opening of pore volume initially
present but inaccessible to helium. It was shown, based on the change in
mercury density, that substantially all of the burning occurred internal to
the particles and that thus the size of the particles did not change significantly
during burning.
The measured total pore volume of a sample was determined by inserting
experimentally measured values of the mercury and helium densities into
equation 1. If the new pore volume resulting from activation was only the
void volume equal to the volume of carbonaceous material removed by re-
action, the total pore volume in a sample would equal the initial pore volume
plus the mass divided by the density of the material removed by reaction. The
density of the removed material is not known nor is it measurable. In order
to estimate the new pore volume created, it was assumed that the reacted
solid had the same density as the remaining solid. A value of 2.0 g/cc was
assumed for the density of the non-porous solid, because that was the highest
helium density measured on any of the activated samples in this work and be-
cause it was the value approached by all samples at high percentage burnoffs.
Thus, 2.0 g/cc seemed to be a good approximation of the true solid density.
51
-------
Total pore volumes calculated in the preceding manner will be called those
"calculated due to burnoff". When these volumes were compared with
measured total pore volumes, it was found that the measured values were
greater than the calculated values in twenty-five of the thirty cases for which
measured total pore volume values were available (the data used were from
Tables 7 and 7a with burnoffs referred to unburned calcinate). Four of the
five exceptions were at burnoffs of 5% or less, where both the measured and
calculated volume changes were very small; the fifth exception was at 51%
burnoff where it is likely that there was some external burning. Thus the
evidence is strongly indicative that the total pore volume increased by more
than would be predicted based on the volume of material removed.
A model for total pore-volume development during burning was derived from
the foregoing observations by assuming: 1) all carbon removal was internal,
and particle size did not change and, 2) that the change in helium density was
due to opening of closed pores. The development of the model is given in
Appendix 1. The model gives the total pore volume, in cc/g of starting
material, after any fraction burnoff, f, by the equation,
(5)
That is, the total pore volume should be predictable from the initial mercury
density, the fraction burnoff and the sample helium density. Values of the
total pore volume in a sample calculated from equation 5 will be called those
"calculated by model".
Figure 12 shows representative values of the total pore volume as measured,
as calculated due to burnoff, and as calculated by model. The values are
plotted as the ratio of the total pore volume of a sample at a certain burnoff
to the initial total pore volume of that sample (at zero burnoff). This ratio
was used because the samples had different initial total pore volumes.
Figure 12 shows that through 15% burnoff the measured and calculated-by-
model total pore volumes agree very well; also that at 30 and 40% burnoff
the measured values fall below the calculated-by-model values but fall above
the values calculated due to burnoff alone. At 51% burnoff the measured
total pore volume was less than that calculated by either method. As indicated
earlier, it is likely that external burning occurred in that sample.
The results shown in Figure 12 lend strong support to the assumptions of the
model. In particular, they tend to confirm the idea that the change in
helium density is due to the opening of closed pores and that these opened
pores contribute to the measurable total pore volume. The rather good corre-
lation between measured and calculated-by-model total pore volumes
suggests that there is not much total pore-volume change due to heating alone,
even at the higher temperatures, in a predominantly carbon dioxide atmosphere,
52
-------
2.0,
I
o
PM
n
-o--
n r r
Measured
Calculated by model
Calculated due to burnoff
o
H
I
c
M
O
O
H
-------
However, thermal pore-volume changes as well as particle-size changes may
account for those differences that do occur between the measured and the cal-
culated-by-model total pore volumes.
Micropore Volume
Figures 5 and 6 show that there is a close correlation between micropore
volume and helium density, at least through 12% burnoff. Based on this
observation, it was postulated that the closed pores that were opened by
burning were micropores. To test this postulate, the burning model was
extended to attribute micropore -volume changes only to the opening of closed
pores. This development, shown in Appendix 1 , gave the result that the
micropore volume, Vpm, in cc/g of starting material, should be given by
- (l-f)/pHe - f/pc . (6)
A "calculated micropore volume'at each burnoff was determined from equa-
tion 6.
The calculated and measured micropore volumes up to 12% burnoff are
shown in Figure 13 (measured micropore volume refers to measured total
pore volume, minus measured macropore volume). It is to be observed
that there is excellent qualitative and quantitative agreement. There is a
slight effect of temperature, with higher temperatures giving higher mea-
sured volumes but giving about the same calculated volumes as lower temper-
ature. This temperature effect is unexplained. Nonetheless, this result
shows that micropore -volume changes, at temperatures up to 400° C and
burnoffs up to 12%, were essentially due entirely to the opening of closed
pores. At higher temperatures and burnoffs, this correlation did not occur.
Thus, it has been shown that pore-volume changes occur by three mechanisms:
the first is due to heating only, which seems not to be important under acti-
vation conditions; the second is due to the void space left by the removal of
carbon; and the third is a result of burning making accessible closed pores
that were already present in the starting material. Both burning and open-
ing of closed pores were shown to contribute to total pore volume when the
burning took place in carbon dioxide at 800° C, and above. Total micropore
volumes calculated from equation 6 are compared with measured total
micropore volumes in Table 11. It is quite clear from the results in Table 11
that the measured and calculated total micropore volumes do not agree for
carbon dioxide activation. Indeed, though the calculated micropore volumes
increased with burnoff, the measured volumes tended to decrease slightly.
54
-------
.25
o
o
o
8
.20
0)
s-l
o
O-,
o
.15
.125
O Measured
Calculated
I
5.
Burnoff,
10.
FIGURE 13. MEASURED AND CALCULATED MICROPORE
VOLUMES VERSUS BURNOFF
55
-------
Table 11 - CALCULATED AND MEASUBED TOTAL MICROPORE
VOLUMES FOR CARBON DIOXIDE ACTIVATION
r
SAMPLE PERCENT
PREPARATION BURNOFF
Carbon diox-
ide, 800° C
C00, 900° C
2t
CO ; 15% burn-
off at 800° C ,
remainder at
900°C
CO2; 30% burn-
off at 800°C,
remainder at
900° C
0
15
30
15
30
27.8
40.5
40.5
51
rOTAL MICROPORE VOLUME, cc/g of
starting material
Measured
.11
.09
.07
.10
.11
.09
.08
Calculated
—
.17
.21
.18
.20
.19
.22
.09
.10
.20
.22
There are several possible explanations for the differences in micropore
results from carbon dioxide activation as compared to those from air acti-
vation. One is that up to 12% burnoff, opened pores were micropores; but
with increased burning, these micropores burned and were enlarged to
become macropores. A second is that as macropores grew by burning,
they grew to include or engulf the existing micropores. A third is that in
carbon dioxide, the opened pores were macropores. None of these expla-
nations is entirely satisfactory.
It is difficult to conceive that quite suddenly, between 12 and 15% burnoff,
the micropores burned to become macropores or that macropores engulfed
micropores. Further, none of the micropore cumulative volume curves
showed significant volume increases in pores greater than about 50 A radius,
suggesting that in no case was there sufficient growth of micropores to be-
come macropores. On the other hand, it is difficult to conceive that pores
that were opened as micropores by air were macropores when opened by
carbon dioxide. It should be remembered, though, that heating alone to
temperatures similar to those necessary for carbon dioxide activation did
cause pore-volume changes, notably increases in macropore volume and de-
creases in micropore volume. Still another observation worth considering
56
-------
was that, whereas the reaction rate with carbon dioxide increased during
the first 10% burnoff (this is ascribed to the exposure of additional reactive
sites), the reaction rate remained constant throughout the first 12% burnoff
with air. This suggests that the opened pores were reactive to carbon diox-
ide but not to oxygen; this is consistent with the growth of micropores upon
carbon dioxide activation. It is unfortunate that there is not more overlap of
the data for the two activating gases. From the available data it is only possi-
ble to observe that the development of micropore volume was quite different
in the cases of oxygen and carbon dioxide as the activating gases; it is not
possible to determine why they are different.
The results in Tables 7 and 9 show that the BET surface area increased with
burnoff. Indeed, from Tables 8 and 9 it can be seen that there is a corre-
lation between the volume of micropores accessible to nitrogen and the BET
surface area, as would be expected since both values are obtained from
nitrogen isotherm data. This correlation is further emphasized in Table 12
and in Figure 14. The plot in Figure 14 shows that surface area is related
to the corresponding volume by a straight line through the origin. Although
there is considerable scatter, the trend is very distinct. The inverse of the
slope of Figure 14 is related to the average pore radius. Assuming cylindri-
cal pores, an average micropore radius of about 6 A is obtained. Table 12
shows that the volume of micropores, less than 10 A radius and accessible to
nitrogen, increases considerably in the first 15% burnoff and is relatively
constant at greater burnoffs.
In the Results section it was pointed out that during carbon dioxide activation,
the rate of reaction in the micropores was partially diffusion limited. There
are not sufficient kinetic data to determine whether diffusion was rate limiting
in air activation. However, since the reaction rates were comparable and the
diffusivities were lower at lower temperature, it seems probable that diffu-
sion was at least partially rate limiting in the pores during air activation.
Nonetheless, partial diffusion control hardly seems an adequate explanation
for the fact that essentially all burning occurred in the macropores. The
surface area in the micropores was always greater than that in the macro-
pores, frequently by a factor of ten. Thus, if the chemical reaction rate
per unit surface area was independent of pore size, then even if diffusion reduced
the rate by 50%, it would be expected that considerable reaction would still
occur in the micropores. That this did not occur is strong, indirect evidence
that the surface is not uniformly reactive, but rather that the surface in the
macropores is more highly reactive than that in the micropores. A possible
explanation for this is that active sites, for example catalytic sites, are not
uniformly distributed. That such is the case seems quite possible, but there
is no known direct evidence that it is true.
57
-------
Table 12 - MICROPORE VOLUME AND SURFACE AREA
SAMPLE
PREPARATION
Starting
material
Carbon diox-
ide, 800° C
CO 900°C
£i
CO ; 15% burn-
off at 800° C, re-
mainder at 900°C
CO -30% burn-
off ^at 800 ^.re-
mainder at 900
-------
70.
0.
.01
.02 .03
Micropore Volume, cc/g
FIGURE 14. MICROPORE SURFACE AREA AS A FUNCTION
OF MICROPORE VOLUME
59
-------
In summary, the observed micropore-growth phenomena are not fully under-
stood. It can be stated, however, that little or no new micropore volume
was created by burning. Increases in accessible micropore volume upon
air activation were due to closed pores originally present in the carbon that
were made accessible by burning. Total micropore volume accessible to
helium tended to decrease upon carbon dioxide activation, while nitrogen
surface area increased. The surface area increases corresponded to in-
creases in pore volume accessible to nitrogen.
Macropore Volume
As shown above, the opening of closed pores accounted for all the micropore-
volume change upon air activation, so the burning must have produced only
macropores. The pore-burning model was extended by assuming that all
burning occurred in macropores and that macropore-volume changes were due
to burning only (see Appendix 1). This yielded for the macropore volume the
equation,
V = V + — (7)
pM pMo p ' v '
(_;
Equation 7 gives calculated macropore volumes. The carbon density, p ,
was taken at 2.0 g/cc. Calculated macropore volumes are compared with
macropore volumes measured by mercury penetration, each plotted against
burnoff in Figure 15 for burnoffs up to 12%. The measured values scatter,
but follow the calculated trend and the scatter about the calculated line is
quite random. The scatter is emphasized by the scale, but is of the order
of the calculated change. The scatter (5. 8% average absolute deviation from
line) may reflect reproducibility of the data. It is concluded that for burning
the calcined coal in air up to 400° C and 12% burnoff, substantially all of the
burning occurs in the macropores, that existing closed pores are opened as
a result of burning, and that the opened pores are micropores.
Since opened pores did not contribute to micropore volume during carbon
dioxide reaction, they must have contributed to macropore volume. That
this in fact did occur may be seen by comparing the values of measured
macropore volume with values that were calculated by attributing both burn-
ing and opening of pores to macropores, as shown in Table 13. The cal-
culated values are from the equation (see Appendix 1) ,
V =V +
pM pMo
The results in Table 13 show that the calculated and measured macropore
volumes agree quite well, except at 51% burnoff where external burnoff
seems to have been important. These results thus demonstrate that upon
60
-------
.35
O
O
° .30
O
d,
O
.25
O Measured
— Calculated due to burnoff
1
5.
10,
Burnoff,
FIGURE 15. MEASURED AND CALCULATED MACROPORE
VOLUMES VERSUS BURNOFF
61
-------
carbon dioxide activation, essentially all the pore volume created by burn-
ing and by opening of closed pores contributed to macropore volume.
Table 13 - MEASURED AND CALCULATED MACROPORE VOLUMES
FOR CARBON DIOXIDE ACTIVATION
SAMPLE
PREPARATION
MACROPORE VOLUME, cc/g of
starting material
BURNOFF
1 Measured
CO0 800° C
2
0
15
30
CO 900 °C
£i
CO ; 15% burn-
off at 800°C, re-
mainder at 900°C
CO ; 30% burn-
off at 800°C, re-
mainder at 900° C
15
30
27.8
40.5
40.5
51
.47
.66
.70
.58
.65
.67
.77
.74
.72
Calculated
.60
.72
.61
.71
.69
.78
.76
.83
Macropore growth is thus rather well understood, resulting almost exclu-
sively from burnoff of carbon at lower burnoffs and from burnoff and opening
of closed pores at higher burnoffs. It might be anticipated that the develop-
ment of the macropore .structure should be predictable.
It was shown earlier that in carbon dioxide activation, the reaction rate in
the macropores was chemical reaction controlled, not diffusion controlled.
This is also almost certainly the case in air oxidation since the rates are
comparable, although this was not explicitly established. One implication
of chemical reaction control is that the gas composition throughout the macro-
pore structure was uniform. It if is further assumed that the reactivity of
the surface was the same in all macropores, regardless of pore size, then
it would be expected that the chemical reaction rate per unit surface area
would be the same in all macropores. With this as the basic assumption, a
model was developed to predict the macropore-size distribution resulting
from burning some fraction of the carbon with a given initial pore-size distri-
bution. The development of the model is shown in Appendix 2. A computer
program was written to do the calculations.
62
-------
For computational convenience, use was made of the fact that the macropore-
size distributions were observed to resemble log-normal distributions. A
method was developed for fitting the macropore-volume distributions with a
log-normal-distribution function. The development of this method is de-
scribed in Appendix 3. The use of this empirical pore-size-distribution
function was not necessary to implement the macropore growth model, but it
made programming the computations much easier.
An example of the results predicted by the macropore burning model, com-
pared with the measured pore volume, is shown in Figure 16 in the form of
normalized cumulative volume curves. The results shown in Figure 16 are
typical; that is, the model did not predict the observed results. Indeed, it
did not even come close to predicting them. This was true for the air-acti-
vated samples as well as for the carbon-dioxide activated samples. Figure 17
perhaps demonstrates even more clearly the differences by showing the pre-
dicted and observed normalized volume distributions. It can be seen that the
actual burning caused the smaller radius pores to grow larger much more
rapidly than the model predicted, resulting in a shift of the distribution's
maximum to much higher values than predicted by the model.
This comparison clearly demonstrates that the assumptions of the macropore-
burning model are not correct. The assumption most likely to cause the dis-
crepancies seen is that of uniform surface burning. Thus, the results strong-
ly indicate that the reaction rate per unit surface area is not the same in all
pores. It is possible that the assumption of cylindrical pores introduces
error, but the model is not especially sensitive to the assumed geometry.
Opening of the mouths of ink-bottle pores might account for the observed
shifts in pore-size distribution. It has been suggested that gas activation
results in reaction primarily at the pore mouth, presumably due to the pres-
ence of catalytic sites there (21, 22).
When the foregoing theoretically-based model was found to fail, several
strictly empirical models were devised to attempt to correlate the results.
Some of these were slight improvements, in particular one in which it was
assumed that the volume change of pores in a given size range was propor-
tional to the initial volume of pores in that range; but none adequately fitted
the data (41).
The results indicate that even in macropores under chemical-reaction-con-
trolled conditions, the rate of reaction per unit surface area was not uniform.
It has already been noted that the surface in the micropores must be much
less reactive than that in the macropores. There is thus considerable evidence
that the reactive sites are not uniformly distributed in pores of different sizes.
It would seem that the nature and distribution of the reactive sites in carbon
are the keys to the development of the carbon's pore structure upon activation.
63
-------
1.0
0.8
£
f
Q)
0.6
O
a.
o
0.4
4-1
cfl
t-l
i
3
0
-a
a)
0.2
0.0
0 % burnoff,
measured
07» burnoff,
predicted
Calcined coal
reacted with carbon
dioxide at 800°C
I
.01
0.1 1.0
Pore Radius, p., log scale
10.
FIGURE 16. PREDICTED AND MEASURED NORMALIZED MACROPORE VOLUMES VERSUS RADIUS
-------
01
01
!>
'0.4
§0.3
•H
4J
O
c
O
•H
0.2
Q
lo,
O
0)
N
cfl
go.o
£ .01
30% burnoff,
predicted
Starting
material
307= burnoff,
measured
0.1 1.0
Pore Radius, MI, log scale
10.
FIGURE 17. PREDICTED AND MEASURED, NORMALIZED, MACROPORE-VOLUME DISTRIBUTIONS VERSUS
RADIUS
-------
Comparison with Other Results
The results of some studies similar to the present one have recently been
published. Turkdogan et al. (21) reacted several different carbons with
carbon dioxide at 1000° C. They found increases in total pore volume and
macropore volume with burnoff. Surface area increased with burnoff up to
about 10 to 20% burnoff and decreased slightly with higher burnoffs. This
they took as evidence of growth of micropores until these grow into one
another thereby destroying surface area. They did not present values of
micropore volume or distribution. These observations are consistent with
those of the present work which indicated micropore growth at low burnoffs
and decreasing micropore volumes at higher burnoffs. They also concluded
that only a small portion of the surface area participated in the oxidation re-
action and that little of this was micropore surface. This is also consistent
with the observations on non-uniform surface reactivity in the present work.
They made no observations with respect to the opening of closed pores.
Chiche et al. (25) studied the pore structures of two carbonized coals acti-
vated with steam at 900° C. They found that surface area from low-angle
X-ray scattering was much greater than nitrogen surface area in the start-
ing materials, indicating a large volume of micropores inaccessible to ni-
trogen. As burnoff proceeded, they found that the X-ray surface area in-
creased little, but the nitrogen surface area increased rapidly, and the two
became essentially the same at high burnoff. This indicates that burnoff
was opening initially closed pores. Indeed, they concluded that the principle
effect of activation was to expose the already existing pore structure. They
also found that pore volume increased with burnoff up to 40 to 50% burnoff
and that most of the pore-volume increase occurred in the micropores rather
than the macropores. Their observations and conclusions are similar to those
of the present study, except for the predominance of micropore over macro-
pore volume growth.
Chiche et al. found some differences in behavior between the two coals. They
also found that samples that had been ground and briquetted behaved somewhat
differently than granular material. The principle difference noted was that
in the granular form of the high rank coal, micropore volume steadily de-
creased with activation. Briquetted samples of this coal, on the other hand,
showed micropore growth. They concluded that the optimal pore structure
for a starting material for activated carbon would be bimodal, containing
micropores to provide surface area and large macropores to provide access
to the interior of the particle. They contend that macropores in the radius
range .5 to 5(j, contribute little to activation in that they burn and consume
carbon, but provide little surface area.
66
-------
Donnett et al. (26) prepared activated carbon from coal by flotation to re-
duce mineral content, followed by partial oxidation by air at 225-325° C
carbonization at 650 °C and then reaction with steam at 900 °C. Throughout
the carbonization step, they found that surface area and micropore volume
measured by carbon dioxide adsorption exceeded those measured by nitrogen
adsorption. They also found that oxygen was driven off, and "crater-like"
pores in the 10 to 500 A radius range were formed by carbonization.
Higher oxygen contents, from higher temperature air oxidations, resulted
in larger crater-like pore and enlarged small micropores. Steam activation
to 50% burnoff resulted in both increased surface area and increased micro-
pore volume, and all the surface area became accessible to nitgrogen. Acti-
vation to 60 to 70% burnoff resulted in little additional change. They did
not report macropore volumes. Their results are in several respects simi-
lar to those of the present study, but differ in that considerable micropore
growth was observed even at high burnoffs.
Singh (58) studied steam activation of an anthracite coal at 830 and 900°C.
He concluded that in the 2. 5 to 12 A diameter range, preheating the coal
to the reaction temperature created new pores and that activation opened
blocked pores (i.e., made smaller pores accessible to larger molecules)
and caused the pores to grow larger. Implicit in his results was the fact
that pores larger than 12 A diameter grew in volume more rapidly than
those smaller.
Marsh and Rand (61) recently reported on studies of the changes in porous
structure of carbon upon gasification. They found that, with a very pure
carbon (polyfurfuryl alcohol carbonized at 850° C) reacted with carbon diox-
ide at 800 °C, there was micropore growth by burning and by the opening of
closed pores. The closed pores were opened during the first 10% burnoff.
The micropore volume per g of starting material increased with burnoff
to a maximum at about 60% burnoff, and then decreased. When iron or nick-
el was added to the carbon, to catalyze the reaction with carbon dioxide,
macropores were formed. There was no micropore growth until after the
metal had become oxidized and its catalytic activity thereby eliminated. They
also observed that carbon monoxide added to the reactant gas enhanced micro-
pore growth in the carbon.
These several studies, plus the present work, have the following results in
common: gaseous activation of carbons has as a major effect the opening,
or making accessible, of micropore volumes already present in the carbon;
pores increase in size due to activation, but growth is not proportional to
surface area (the surface is not uniformly reactive to the reactant gas with
macropore surface generally appearing to be more reactive); activation in-
creases micropore volume and surface area at low burnoffs, but as burnoff
increases, micropore volume and area tend to be relatively constant or even
decrease. The most substantial difference between the observations of the
67
-------
present work and those of other investigators was that in the present study
there was essentially no burning in the micropores, while others have found
that there apparently was burning in the micropores.
Putting all these results together, several general observations about acti-
vation and pore structure can be made. The pore structure of a porous
carbon is affected by heating in inert gases. The effect tends to be one of in-
creased total and macropore volumes and decreased micropore volume. The
gas is important; greater pore volume changes occur in helium than in nitro-
gen. The burnoff of material tends to be non-uniform, with far more burnoff
per unit area occurring in macropores than in micropores. In some cases
there is little or no burning in micropores, while in other cases there appears
to be substantial burning in micropores. In any case, the principle effect of
activation upon micropores, and thus upon surface area, is to make access-
ible micropores already present in the starting material. The relatively
lower reactivity of micropore surface than macropore surface may be due to
diffusion-limited reaction in the micropores, but more probably it is due to
a non-uniform distribution of reactive sites. Indeed, there appears to be a
non-uniform distribution of reactive sites even within the macropores. The
exposure and growth of micropores and surface area occur mainly at low burn-
offs (usually less than 20%), and at increased burnoffs, micropore volume
and surface area remain relatively constant or even decrease.
It is interesting to examine the results of a recent study on gaseous reactiva-
tion of carbons used in wastewater treatment in the light of the present knowl-
edge of the activation process. Juhola and Tepper regenerated spent acti-
v ated carbon from the Pomona, California tertiary sewage treatment plant
(13). They found that once-spent carbon could be regenerated to give essen-
tially the same properties as the virgin carbon, but that carbon spent at
least twice and regenerated at least once could then not be regenerated to
give the properties of the virgin carbon. In particular, they found that al-
though the mercury and helium densities and the molasses number could be
matched with those of the virgin carbon, the iodine number was lower. This
indicates that the carbon contained fewer pores in the 5 to 14 °A radius range.
They also pointed out that a phase of the reactivation process, which they
called baking, involved carbonization of adsorbed organic materials. It was
observed in the present study that little or no reaction occurred in the micro-
pores, which include those that are measured by the iodine number but not by
the molasses number. Thus, it is proposed that carbonization deposits car-
bon on pore surfaces and that regeneration removes little, if any, of the de-
posited carbon from the micropores. Successive use and regeneration of
activated carbon would thus result in an accumulation of carbon in the micro-
pores and a progressive decrease in micropore volume and hence in iodine
number. This explanation suggests that, unless some way could be found to
make the carbon in the micropores more reactive, it should not be expected
68
-------
that regeneration would recover the micropore volume, and hence iodine
number and surface area, of the virgin carbon.
69
-------
SECTION VH
ACKNOWLEDGEMENTS
This report was written by R. E. West, Associate Professor of Chemical
Engineering, University of Colorado. Associate Professor L. F. Brown
directed major portions of this work and contributed throughout the project.
Much of the early initiative for this project was provided by Dr. J. H. Blake.
Dr. W. Manogue made important, helpful suggestions.
The experimental studies and analysis of the results were conducted by
graduate students as theses and special projects, and their contributions are
gratefully acknowledged. The students were S. M. Fass, W. M. Kalback,
S. N. Sharma, R. S. Parekh, G. V. Desai, M, M, El Tawil, P. W. Allen, and
S. V. Mangu.
Important assistance in equipment construction and maintenance was provided
by Messrs. D. Page and E. Tomazewski of the Chemical Engineer Laboratory.
Some of the material used in this project was donated. The cooperation of
the FMC Corporation, Pure Carbon Company, North American Carbon, Inc.,
and Chemtron Chemical Corporation is gratefully acknowledged.
The financial support of the Environmental Protection Agency made this work
possible. Their support is gratefully acknowledged. Financial support to
some of the students was provided through the Esso Education Foundation, the
Dow Chemical Co. , the Monsanto Co. , and the University of Colorado.
The cooperation and patience of Mr. Robert H. Wise, EPA Project Officer,
is especially appreciated.
71
-------
SECTION VIII
REFERENCES
1. Parkhurst, J.D. , F.D. Dryden, G.N. McDermott, and J. English,
J. Water Poll. Control Fed , 39, R70 (1967).
2. Slechta, A. F. , and G. L. Culp, J. Water Poll. Control Fed. , 39,
787 (1967).
3. Smith, D. R. , and H. F. Berger, J. Walter Poll. Control Fed. , 40,
1575 (1968).
4. Weber, W.J. , Jr., C. B. Hopkins, and R. Bloom, Jr., J. Water Poll.
Control Fed. , 42, 83 (1970).
5. Johnson, R. L. , F. J. Lowes, Jr., R. M. Smith, and T.J. Powers,
U.S. Public Health Service, Report No. AWTR-11 (1964).
6. Bishop, D. F. , L. S. Marshall, T. P. O'Farrell, R. B. Dean, B. O'Conner,
R.A. Dobbs, S.H. Griggs, and R. V. Villiers, iJM§LZ9lL_Control
Fed. , 39, 188 (1967).
7. Mantell, C. L. , Carbon and Graphite Handbook, Interscience Pub. (1968).
8. Carlton, S. S. , C. M. Marr, Jr., C.W. Foye, C.K. Owens,
R. A. Chiancone, and B. C. Raynes, "Investigation of the Use of Coal for
Treatment of Sewage and Waste Waters", Final Report, Contract
14_Ol-0001-348, Office of Coal Research, Dept. of Interior, Wash. ,B.C.
(1966).
9. Shannon, E. , and P. L. Silveston, Chem. Eng. Prog. Symp. Series. 64,
# 90, 198 (1968).
10. Brunauer, S. , P. H. Emmett, and E. Teller, J. Am. Chem. Soc. . 60,
309 (1938) .
11. Mattson, J. S. , L. Lee, H. B. Mark, Jr., and W.J. Weber, Jr.,
J. Colloid Interface Sci. . 33, 284(1970).
12. Brunauer, S. , Chem. Eng. Prog. Symp. Series. 65, #96, 1(1969).
13. Juhola, A.J. , and F. Tepper, Robert A. T aft Water Research Center,
Report No. TWRC-7. Feb. (1969).
73
-------
14. Snoeyink, V. L. , and W. J. Weber, Jr., Advances in Chemistry Series.
Vol. 79, 112 (1968).
15. Scott, D. S. , and F. A. L. Dullien, AIChE_Journal, 8, 293 (1962).
16. Hsieh, J. S., R. M. Turian, and C. Tien, Research Report 69-2,
Grant No. 17020 DZO, Federal Water Pollution Control Administration
(1969).
17. Hill, A. , and H. Marsh, Carbon. 6, 31 (1968) .
18. Dubinin, M. M. , Proc. Fourth Int. Symp. on Reactivity of Solids.
J.H. deBoer et al. , eds., 643, Elsevier (1961).
19. Cameron, A., and W. O. Stacey, Australian J. ,Appl. Sci.. J), 283
(1958).
20. Walker, P. L., Jr., and E. Raats, J. Phys. Chem. . 60, 364(1956).
21. Turkdogan, E.T., R. G. Olsson, and J. V. Vinters, Carbon. 8, 545
(1970).
22. Kadlec, O., A. Varhanikoon, and A. Zukal, Carbon. _8, 321 (1970).
23. Lamond, T. G. , and C. R. Pierce, J. Colloid Interface Sci. . 31,
104 (1969).
24. Dubinin, M. M. , and G. M. Plavnik, Carbon, £, 183 (1968).
25. Chiche, P., J. Coue, S. Durif, and S. Pregermain, Carbon 1, 297
(1969).
26. Donnet, J. B., P. Couderc, L. Kobel, and E. Papirer. Carbon, j3,
63 (1970).
27. Doying, E.G., Encyclopedia of Chemical Technology, R. E. Kirk, ed. ,
v. 4, 149, John Wiley and Sons (1964).
28. Allen, P.W. , Ph.D. Thesis, University of Colorado (1968).
29. Dullien, F. A. L. , and V. K. Batra. Ind^JEng^jChem, , 62, 25(1970).
30. jlandbook of Chemistry and Physics, 49th edition. R. C. Weast, ed. ,
F-151, Chemical Rubber Co. (f968).
74
-------
31. van der Plas, Th. , Physical and Chemical Aspects of Adsorbents and
Catalysts. B.C. Linsen, ed. , 425, Academic Press (1970).
32. Numinco Corp. , Operating Manual, Numinco-Orr Surface Area-Pore
Volume Apparatus, Model No. MIC-101.
33. American Instrument Co. , Operating Manual, Aminco-Winslow Mercury
Porisimeter, Model No. 5-7108.
34. Emmett, P. H. ,' Catalysis, v. 1, P.H. Emmett, ed. , 32, Reinhold
(1954).
35. Fass, S. M. , Ph.D. Thesis, University of Colorado (1967).
36. Orr, C. , Jr., and J. M.' Dallavalle, Fine Particle Measurement. 267.
MacMillan (1959).
37. Ritter, L. C. , and R. L. Drake, Ind. Eng. Chem. , Anal. Ed. . 17,
782 (1945).
38. Joyner, L. G. , E. P. Barrett, and R. Skold, J. Am. Chem. Soc. , 7_3,
3155 (1951).
39. Sharma, S.N. , M.S. Thesis, University of Colorado (1967).
i
40. Kalback, W. M. , M.S. Thesis, University of Colorado (1967).
41. Parekh. R. S. , M.S. Thesis, University of Colorado (1968).
42. Desai, G. V. , M.S. Thesis, University of Colorado (1968).
43. El Tawil, M. M. , M.S. Thesis, University of Colorado (1970).
44. Hartley, H. O. , Techno metrics, 3, 269 (1961).
45 Cutlip, M. B. , Ph.D. Thesis, University of Colorado (1968).
46 Roberts G. W. , and C.N. Satterfield, Ind. Eng. Chem. Fundamentals^
4, 288 (1965).
47. Wakao, N. , S. Otani, and J. M. Smith, AIChE Journal. 1, 435 and
439 (1965).
48. Walker, P. L., Jr., F. Rusinko, Jr., and E. Raats, J. Phys. Chem.,
J59, 245 (1955).
75
-------
49. Blake, J. H. et al. ; paper presented at the 147th National Meeting of the
American Chemical Society, Philadelphia, Pa., April (1964).
50. Ergun, S., Kinetics of the Reaction of Carbon Dioxide and Steam with Coke,
Bureau of Mines Bulletin 598 (1962).
51. Walker, P. L., Jr., F. Rusinko, Jr., and L. G. Austin, Advances in
Catalysis, XI, 133, D. D. Eley et al. f eds., Academic Press (1959).
52. Thomas, J. M, , Carbon. 8, 413 (1970).
53. Atkins, J.H., Carbon. 3, 279 (1965) .
54. Kalback, W. M., L. F. Brown, and R. E. West, Carbon. JJ, 117 (1970).
55. Dubinin, M. M. , G. M. Plavnik, andE.D. Zaverina, Carbon, J2, 261
(1964).
56. Walker, P. L., Jr., and F. Rusinko, Jr., J. Phys. Chem. . 5J), 241
(1955).
57. Hennig, G. R., Chemistry and Physics of Carbon, P. L. Walker, Jr.,
ed., J, 1, Dekker (1966).
58. Singh, D.D., Chem. Ind. (London^, 1968, 620 (1968).
59. Miller, I., and J.E. Freund, Probabilitv and Statistics for Engineers
77, Prentice-Hall. Inc. (1965).
60. Mangu, S. V., unpublished report, Chemical Engineering Department,
University of Colorado (1970).
61. Marsh, H., and B. Rand, Carbon, 9, 47, 63, 79 (1971).
76
-------
SECTION IX
LIST OF PUBLICATIONS
1. Kalback, W. M. , L. F. Brown, andR.E. West, "The Growth of Pores
in Graphitized Carbon Reacted with Carbon Dioxide", Carbon, 8.,
117-124 (1970).
2. Fass, S. M. , "The Effect of Reaction Mode on the Development of Pore
Structure in Active Carbon", Ph.D. Thesis, University of Colorado
(1967).
3. Kalback, W. M. , "Pore Growth in Graphitized Carbon Reacted with
Carbon Dioxide and Diffusion Limited Conditions", M.S. Thesis,
University of Colorado (1967).
4. Sharma. S. N. , "The Transition Zone Behavior of the Carbon-Carbon
Dioxide Reaction", M.S. Thesis, University of Colorado (1967).
5. Allen, P.W. , "Unsteady State Diffusion in Porous Media", Ph.D.
Thesis, University of Colorado (1968).
6. Parekh, R. S. , "The Effect of Carbon Monoxide on Pore Development
in Active Carbon", M.S. Thesis, University of Colorado (1968).
7. Desai, G. V. , "Kinetics of Carbon-Carbon Dioxide Reaction: Gasification
of a Calcined Coal", M.S. Thesis, University of Colorado (1968).
8. El Tawil, M. M. , "Mechanism of Pore Structure Development During
Air Activation of Carbon", M.S. Thesis, University of Colorado (1970).
77
-------
SECTION X
GLOSSARY
Terms and Abbreviations:
Activated Carbon - Any form of carbon with large pore volume and surface
area capable of adsorbing gases or liquid solutes.
Activation - The process of reacting a carbon, usually with an oxidizing gas,
to produce activated carbon.
BET Surface Area - Surface area of a porous solid determined from gas ad-
sorption, usually nitrogen, by the method of Brunauer, Emmett and Teller.
Burnoff - Fraction or percentage of the mass of starting material removed
by reacting with oxidizing gas.
Calcinate - Carbon prepared by heating in inert atmosphere to 600 - 800°C.
Carbonization - Devolatilization and cracking of adsorbed hydrocarbons which
leaves carbon.
Char - Carbon prepared by heating in inert atmosphere to temperatures up
to ca. 500"C.
Chemical rate controlled - Reaction can occur no faster than the inherent
chemical reaction rate.
o
Crystallites - Small crystals of graphitic structure from perhaps 5 to 2000 A
in thickness.
Diffusion controlled or limited - Reaction can occur no faster than a rate de-
termined by the rate of diffusion of reactants or products.
Helium density - Density of solid carbon determined by helium displacement
which is the mass of carbon divided by volume not penetrated by helium.
o
Macropores - Pores with radii greater than 100 A.
Macropore volume - Volume determined by mercury penetration which
includes pores with radii between 100 A and 51 M- .
79
-------
Mercury density - Density of solid carbon determined by mercury displace-
ment at ca. 1.7 psia, or mass of carbon divided by the sum of volume of
carbon and volume of pores.
b
Micropores - Pores with radii less than 100 A.
Micropore volume - Volume of micropores, determined as total pore volume
minus macropore volume.
Pores - Internal voids in a carbon particle.
Porosity - Total pore volume divided by total particle volume (carbon plus
pores).
Porous structure - Measurable properties of porous carbon such as pore
volumes, volume distribution, and surface area.
Starting Material - Material to be reacted at zero burnoff, but after loss of
volatiles at reaction temperature.
STP - Standard temperature and pressure: 0°C, 1 atm.
Total pore volume - Void volume penetrable by helium but not by mercury, or
reciprocal of mercury density minus reciprocal of helium density.
USS mesh - United States standard sieve sizes.
Symbols:
A. - Preexponential factor in rate coefficients where the subscript number
corresponds to that of rate coefficient: A in % C/hr-atm; A in atm" ;
A in atm .
O
o -10
A - Angstrom unit, 10 meters.
3 - Contact angle of liquid nitrogen in degrees.
dV. - Pore volume in an increment of radius, expressed as cc/g .
2
dS. - Surface area in an increment of radius, expressed as m /g
e - Porosity, a dimensionaless quantity.
AE. - Activation energy of rate coefficients, expressed as kcal/g mol.
80
-------
f - Fraction burnoff, a dimensionless quantity.
f(z) - Normal probability density function, a dimensionless quantity.
F(z) - Cumulative normal probability function, a dimensionless quantity.
i - Subscript which refers to property in an increment of radius.
k. - Rate coefficients in equations :k in %C/hr atm; k in atm ; k in atm .
1 .L £ o
In - Natural logarithm, a dimensionless quantity.
L. - Length of pores in radius increment.
m - Meters, or mass of carbon in g.
p. - Micron, or mean of distribution function.
o - Subscript which refers to property of starting material.
p - Pressure in atm or psia, or subscript indicating pore property.
p - Vapor pressure in atm or psia.
Partial pressure of carbon monoxide in atm.
- Partial pressure of carbon dioxide in atm.
o
r - Pore radius, in either A or p, .
r. - Average radius of a radius increment.
r - Kelvin radius.
K
R - Universal gas constant or reaction rate, expressed as %/hr
p - Density of solid carbon in g/cc.
L;
p - Helium density in g/cc.
rHe
p^ - Mercury density in g/cc.
Hg
2
S - Surface area in m /g.
a - Surface tension in dynes/cm, or standard deviation of distribution.
81
-------
0
t - Thickness of adsorbed film in A , or time in hr .
T - Temperature in K.
9 - Contact angle of mercury in degrees.
v - Specific volume of liquid nitrogen in cc/g.
V - Volume in cc/g .
VQ- Volume of solid carbon, expressed as cc/g of starting material.
V ' - Volume of solid carbon after burnoff, expressed as cc/g of starting
material.
V - Volume of solid carbon in cc/g.
O
V - Volume of closed pores opened by reaction, expressed as cc/g of
starting material.
V - Volume of closed pores not opened by reaction, expressed as cc/g
of starting material.
V - Volume of pores created only by burning of carbon, expressed as cc/g
of starting material.
V - Constant in normalized pore volume, expressed as cc/g of starting
material.
V - Total pore volume in cc/g.
V - Total pore volume in cc/g of starting material.
V - Micropore volume in cc/g of starting material.
V - Macropore volume in cc/g of starting material.
V - Constant in normalized pore volume, expressed as cc/g of starting
material.
x - Dummy variable in distribution function.
z - Any random variable of a normal distribution.
82
-------
SECTION XI
APPENDICES
Page
MODEL OF PORE-VOLUME CHANGE 84
MACROPORE BURNING MODEL 90
FITTING OF MACROPORE DISTRIBUTIONS 93
83
-------
APPENDIX 1: Model of Pore -Volume Change
A method is developed herein for computing the expected change in total,
macropore, and micropore volumes.
It is assumed that the total volume of a carbon particle (that is, the volume
of carbon plus the volume of pores as determined by mercury density) does
not change with burnoff. Or equivalently, that all mass removed by burning
is removed from the interior surface. The mercury density of the starting
material is thus given by
»Hgo - ^TTp ™
The mercury density of the material after burnoff is given by
+V
p
Dividing equation 1-2 by 1-1 gives
The total measurable volume of a sample, in cc/g of starting material,
will be that of the starting material plus the change in pore volume;
V = V + A V (1-4)
p po p v '
The initial pore volume is given by
AV is the change in the total pore volume due to activation and can be
P
expressed as follows:
- 1 1-f
AV = - - - (1-6)
P PHeo PHe
The first term, 1/PH , is the volume per unit weight of all material impen-
etrable by helium in the starting material, and (l-f)/p is the volume im-
penetrable by helium in a sample in cc/g of starting material. The
84
-------
difference between these two quantities gives the change in the volume impen-
etrable by helium. If the particle size has not changed, this quantity will
be the change in interior pore volume. It is to be emphasized that this method
includes the assumption that the particle volume does not change by activation.
Combining equations 1-4, 1-5 and 1-6 results in
v = -_ - (1_7)
P "Hgo "He
V determined in this manner is referred to as the "calculated-by-model"
total pore volume in cc/g of starting material.
It is to be noted that this calculated pore volume makes use of experimental
data. However, it is used to calculate the expected pore volume from the
mercury density of the starting material and from the helium density of the
sample, and it assumes a constant particle volume. Hence, since it uses
a property of the starting material, an assumption about the particle size
and only one measured property of a sample, it is more predictive than the
measured pore volume as given by equation 1-5.
Micropores are considered to be all pores less than . 01^ in radius. The
micropore volume has been obtained by subtracting the measured macropore
volume from the measured total pore volume.
It was noted in the experimental results that the micropore-volume curves
have a similar shape to those of helium densities, suggesting a relationship
between the two. As was pointed out earlier, the change in helium density
is considered to be due to the opening of closed pores. This suggests that
changes in micropore volume may be due to the opening of closed pores. To
examine this possibility, a model will be developed for predicting micropore
volumes as a result of opening closed pores. Results predicted by the model
will then be compared with the actual experimental results.
Figure 18, a and b, will be used to illustrate the changes occurring.
Figure 18a represents a particle of the starting material, and Figure 18b
represents the same average particle after activation. A basis of one g
of starting material will be taken. After burning, the weight of the same
particles will be (1-f) g, and the weight of each particle will be assumed
to be in that same ratio to its initial weight.
85
-------
(a)
(b)
FIGURE 18. SCHEMATIC DIAGRAM OF CARBON PARTICLE
86
-------
The mercury and helium densities can be expressed as follows:
starting material,
weight of carbon, g 1
Hgo Totalvolumeofparticle.ee V_ + V +V +v (1-8)
C p cpl Cp2 '
_ _ weight of carbon, g _
Heo vol. of solid carbon +any closed pores, both in cc
(1-9)
V + V , + V
C cpl cp2
sample after reaction,
weight of carbon remaining, g
p = - *•* - ***-*• — ^
Hg total volume of particles, cc
1-f
V' + ¥„ + V + V , + V (1 10)
f p cpl cp2
weight of carbon remaining, g
He volume of carbon remaining + any closed pores, both in cc
" V' +V v
C cp2
Assuming that there is no significant change in the volume of the particle
due to carbon burnoff. V can be determined by dividing the fraction
burned off by the solid-carbon density. In these calculations, the density
of solid carbon was assumed to be 2.0 g/cc.
To calculate the volume change due to opening of closed pores, the following
steps will be taken:
subtract 1-11 from 1-9 to give.
T - v' + V = v+ V , (1-12)
C C cpl f cpl v ;
87
-------
This expression is the pore-volume change due to both burn-off and opening
of closed pores. Therefore, from 1-12,
— 1 - —
V — - ' — — "~ — V. (1 ~" 1 «5 )
Cpl
The carbon and hence the pore -volume change due to burnoff is,
— f f
(1-14)
2.0
Thus, the volume of closed pores opened as a result of burning is given by,
L-JL -L
(
Assuming that all the closed pores opened up are micropores and that no
burning occurs in the micropores, the change in volume of micropores
(in cc/g of starting material) will be given by equation 1-15, and the
volume of micropores will be given by ,
V = V + V = V + — - - -- - -— (1-16)
pm pmo cpl pmo pReo pHg 2.0
If all the burning occurs in the macropores and the macropo re -volume
change is due only to burning, the macropore -volume change will be given
by equation 1-14, and the macropore volume (in cc/g of starting material)
will be given by ,
o 20
If the macropore-volume change is due to opening of closed pores as well as
to burning, the total change in pore volume, given by equation 1-6, must
be added to the initial macropore volume to give,
>> + -
It is interesting to note that if the change in micropore volume (given by
1-15) and the change hi macropore volume (given by 1-14) are added to the
initial total pore volume (given by 1-5), the result is 1-7 which is the ex-
pression for total pore volume developed above. Thus, the equations are
consistent.
88
-------
APPENDIX 2: Macropore Burning Model
Two different mathematical models are developed to describe the change in
the cumulative macropore volume due to burning. The first, model 1, is
based on the assumption of uniform reaction rate per unit surface area through-
out the macropores. The second, model 2, is based on assuming that the
reaction rate in a pore-size increment is proportional to the initial pore
volume present in that increment.
Model 1
For cylindrical pores of average radius r. and length L., in a radius
increment d£nr the volume dV. and the surface area dS. are given
, i i
by ,
dV. = Trr2L. (2-1)
111 '
dS. = 2-nr.L. (2-2)
111 '
2dV.
dS. = — (2-3)
Dividing 2-2 by 2-1,
A normalized, pore-volume, distribution function is defined by,
d(V -V )/(V -V )
- - ~ <2-4)
Hence ,
dV. - (V - Vo)f(z)dlnr (2-5)
Substituting 2-5 into 2-3 gives for the surface area in the increment,
2dVi 2
dS. = - ~ = ^— (VT - Vo)f(z)dlnr (2-6)
i i
If m is the initial weight of carbon and m is the weight of carbon left
o
after time t, then the rate of reaction, R, is given by ,
89
-------
d(m/m
dt m dt
o
The change in carbon mass, dm, is p dV which is the carbon density
O Vy
times the change in volume of carbon. The change in carbon volume is -dV ,
the negative of the change in pore volume. Substituting this into 2-7 and
dividing both sides by the total initial surface area, S , gives,
,2 8!
(2"8)
S " S m dt
O O O
Now, assuming that the reaction rate per unit surface area is a constant, the
reaction rate in pores of average radius r. is given by,
R. = ~~ dS. (2-9)
i S i v '
o
and substituting for dS. from 2-6 yields,
R. = g— (2(VT - VQ)f (z)cUnr) (2-10)
o i
From 2-8 the rate in the pores is also given by,
<2-n>
Pcd(dV.)
Equating 2-10 and 2-11, rearranging and integrating gives
dV. t
dV C o i 0
io
Rdt (2-12)
90
-------
Integrating equation z-7 gives,
t m
m - m
_ = f 2_
v
. m m
0 mo o
o
Thus, from 2-12, the change in volume of pores of initial average radius
r. is given by ,
f(2(V - VJf(z)cUnr)
A(dV.) = — (2-14)
KC o i
As burning occurs on the pore surface, the radius will also change. If r.
is the mean pore radius in an increment at time zero and r is the mean
radius at time t , the change in volume of the pores is given in terms of
radius by,
2-rrLt
Solving for r.
2 i
+-—— (2-16)
1 Vc
Using equations 2-14 and 2-16, the new volume and radius corresponding to
any increment of the volume distribution of the starting material could be
calculated as a function of burnoff. A computer program was written to do
the calculations . A volume distribution function for the starting material
is needed. This could be in numerical form, as the actual raw cumulative
pore-volume data, or in functional form as described in Appendix 3.
91
-------
Model 2
Assuming that the pore-volume change is directly proportional to the initial
pore volume in an increment, equations are derived to predict the pore-
volume distribution for a sample of a given burnoff. It is to be noted that
this assumption is not made on theoretical grounds, rather as an empirical
observation.
The total macropore-volume change due to burning is given by equation_l-13.
The change in volume in any increment, AdV. , is proportional to dV. and
is equal to the total change in pore volume times the fraction of the total pore
volume in the increment , thus,
fdV.
AdV = ^—=- <2-17>
1 Pc
-------
APPENDIX 3: Fitting Macropore Distributions
As was noted in the Results section, the macropore—distribution curves
were found to resemble the log-normal density function. The log-normal
density function is given by the expression (59)
f(z) = —i- exp ( - |((inx - ^)/a)2) (3-1)
and the cumulative value is given by (59)
z
F(z) = —— exp (-((inx-^)/a)dinx (3-2)
Values of F(z), the standard cumulative normal distribution, are tabulated
(59) or may be calculated by common computer subroutines. In the present
work, F(z) corresponds to the normalized cumulative macropore volume,
(V - V )/(V - V ) . The density function, f(z), corresponds to the
normalized volume -distribution function, (d(V - V )/(V - V ))/ d in r .
In equations 3-1 and 3-2, x corresponds to r, and ^ and a are the
mean and standard deviation of the distribution function. Thus,
f(z) = — exp ( - ((4n r -
Jzv 2
= /d£nr (3-3)
and, Jfor
1 1 ?
F(z) = / — - exp( --((in r - P,)/CT) ) d in r
-oo A/2TT Q
- Vo>
93
-------
A non-linear least-squares fitting program was used to take mercury-
porisimeter data in the form of cumulative volume at various pore radii_
and to obtain from that the best values of the constants u, a Vm and V
r-i , rp, Q
(44, 45, 60). With these values, equation 3-3 could be used to calculate
the normalized volume-distribution curve, and equation 3-4 could be used
to calculate the normalized cumulative-volume curve. The calculated
cumulative-volume curves fit the data with an average absolute error of
2% (60) .
94
-------
1
Accession Number
w
2
Subject Field & Group
05D
SELECTED WATER RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
Organization
Colorado University, Boulder, Colorado
Chemical Engineering Department
Title
EFFECT OF POROUS STRUCTURE ON CARBON ACTIVATION
10
Authors)
West, Ronald E.
16
Project Designation
EPA, WQO Grant No. 17020 DDC
Note
22
Citation
23
Descriptors (Starred First)
*Activated Carbon, *Coals, Pores, Oxidation, Carbon, Chemical
Reactions, Diffusion, Kinetics, Physical Properties
25
Identifiers (Starred First)
*Adsorbents. Carbon Activation,, Porous Structure
27
Abstract
Reaction rates and porous structures of a calcined Wyoming coal activated by air
and by carbon dioxide and a graphite activated by carbon dioxide were measured. Total,
macropore and micropore volumes, surface area and pore-size distributions were
determined as functions of burnoff.
In 1-mm-diameter calcined coal particles, the overall reaction rate was chemical
rate limited at temperatures as high as 961° C and rates as high as 106%/hr. In
19-mm-diameter graphite cylinders, the rate at 1030°C was diffusion limited. In
graphite, pore growth, by end-burning of crystallites was an important mechanism.
In calcined coal, pore-volume changes occurred due to heating, carbon burnoff and the
opening of closed pores present in the starting material, Burnoff of carbon occurred
essentially entirely in macropores. Up to 12% burnoff, the closed pores that were
opened were micropores; above 12% burnoff they were macropores.
In order to develop large micropore volumes and hence large surface areas by gas
activation, a micropore structure must be present in the starting material.
The report includes 13 tables, 18 figures and 61 references. (West-Colorado)
Abstractor
West
Institution
Colorado University
WR-102 (REV. JULY
WRSI C
IUMENT, TO: WATER RESOURCES SCIENTIFIC INFORMATION CENTER
U.S. DEPARTMENT OF THE INTERIOR
WASHINGTON, D. C. 20240
GPO: 1970 - 407 -691
6U.S. GOVERNMENT PRINTING OFFICE: 1972 484-483/89 i-j
------- |