United States
Environmental Protection
Agency
Municipal Environmental Research
Laboratory
Cincinnati OH 45268
EPA-600/2-80-123
August 1980
Research and Development
Preparation and
Evaluation of
Powdered Activated
Carbon from
Lignocellulosic
Materials
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2, Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring ;
5. Socioeconomic Environmental Studies
6, Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports ;
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-80-123
August 1980
PREPARATION AND EVALUATION OF POWDERED ACTIVATED CARBON
FROM LIGNOCELLULOSIC MATERIALS
by
Paul V. Roberts, Douglas M. Mackay, and Fred S. Cannon
Department of Civil Engineering
Stanford University
Stanford, California 94305
Grant No. EPA-R-803188
Project Officer
Richard Dobbs
Wastewater Research Division
Municipal Environmental Research Laboratory
Cincinnati, Ohio 45268
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Municipal Environmental Research
Laboratory, U.S. Environmental* Protection Agency, and approved for publica-
tion. Approval does not signify that the contents necessarily reflect the
views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or recom-
mendation for use.
ii
-------
FOREWORD
The U.S. Environmental Protection Agency was created because of increas-
ing public and government concern about the dangers of pollution to the health
and welfare of the American people. Noxious air, foul water, and spoiled land
are tragic testimonies to the deterioration of our natural environment. The
complexity of that environment and the interplay of its components require a
concentrated and integrated attack on the problem.
Research and development is that necessary first step in problem solu-
tion; it involves defining the problem, measuring its impact, and searching
for solutions. The Municipal Environmental Research Laboratory develops new
and improved technology and systems to prevent, treat, and manage wastewater
and solid and hazardous waste pollutant discharges from municipal and commun-
ity sources, to preserve and treat public drinking water supplies, and to
minimize the adverse economic, social, health, and aesthetic effects of pollu-
tion. This publication is one of the products of that research and provides a
most vital communications link between the researcher and the user community.
The project reported here evaluated the technical feasibility of con-
verting a solid waste (prune pits) into adsorbents suitable for wastewater
treatment.
Francis T. Mayo, Director
Municipal Environmental Research
Laboratory
iii
-------
ABSTRACT
This research project was conceived as a preliminary evaluation of the
technical feasibility of converting solid wastes into adsorbents suitable for
wastewater treatment. The work emphasized the pyrolysis of solid wastes rich
in organic constituents, mainly agricultural wastes. The char prepared from
one of these materials (prune pits) was subsequently activated for comparison
with activated carbons that are widely used in water and wastewater treatment.
Experiments were conducted in laboratory equipment using milligram quantities
of solids.
The char yield from pyrolysis depends^ on educt composition, temperature
and heating rate. For a given pyrolysis temperature, maximum char yield is
attained with educts of high lignin content and low heating rates.
The chars so prepared showed specific surface areas of 300 to 650 m /g,
measured by CO^ adsorption (195K), but the pores were so small that the solids
were penetrated only slowly by No. Activation with COo at 900°C for 30 to 60
min greatly increased the specific surface area, the pore volume, and the size
of pores. Activated carbon prepared by exposing prune pit chars to an atmo-
sphere of C02 at 900°C for 30, 42, and 60 min had surface areas of 930, 1180,
and 1690 m /g (N2, 77K), respectively.
The activated carbons made from prune pits demonstrated favorable adsorp-
tion performance when compared with an activated carbon widely used in water
and wastewater treatment. The prune pit char activated at 60 min demonstrated
a higher adsorption capacity and superior adsorption kinetics than did a
ground commercial product (Filtrasorb 400), when judged according to the up-
take of dissolved organic carbon (DOC) from secondary effluent. An adsorbent
made by activation of prune pit char for 42 min was approximately equivalent
to Filtrasorb 400 in every respect: specific surface, pore size distribution,
adsorption'capacity, and adsorption rate.
The uptake of DOC from secondary effluent by powdered activated carbon
behaved according to a model that assumes linear equilibrium and rate control
by pore diffusion. The apparent diffusivities estimated from the uptake rate
were in the range of 1 x 10~10 to 3 x 10 m /s, conforming to expectations
based on molecular diffusion of organic substances of the sort expected in
secondary effluent.
Samples of refuse-derived fuel (a shredded, organic-rich fraction of
municipal solid waste) were pyrolyzed and activated under the same conditions
as for prune pits. The specific surface area of the activated material from
refuse-derived fuel was only one-fourth that of the corresponding material
from prune pits.
iv
-------
This report was submitted in fulfillment of Grant No. R-803188 by the
Department of Civil Engineering, Stanford University, under the sponsorship of
the U.S. Environmental Protection Agency. This report covers the period
1 November 1976 to 29 October 1979, and work was completed 29 October 1979.
v
-------
-------
CONTENTS
Foreword ........... , ................. ... ill
Abstract ................................ iv
Figures ......... .............. ; ...... . . ix
Tables . ................. . . ...... . . ..... xi
Acknowledgments ........ . ..... ..... . ....... . xiii
1. Introduction ......................... . . 1
Statement of the Problem .................. 1
Objectives .............. ......... . . 1
Research Approach ...... . .............. 2
2. Conclusions ......... ...... ............ 3
3. Recommendations . . .......... ............ . 5
4. Physical Characterization and Preparation of Activated Carbon . . 7
Physical characterization ..... .... ........ 7
Pyrolysis . . ..... .......... . ..... . . 13
Activation ...... ................... 15
5. Measures of Sorptiori Performance ..... ............ 18
Equilibrium isotherms .......... ...... . . . 18
Kinetics of adsorption ...... ........ ..... 20
Adsorption of organics from wastewater effluent ..... . 22
6. Materials and Methods ..... ...... ..... .....'. 24
Cellulose .... ...... . ............ . . 24
Lignocellulosic materials ...... ........... 24
Refuse-derived fuel ................. ... 25
Materials preparation ................... 25
vii
-------
CONTENTS (continued)
6. (cont.)
Raw materials analysis 26
Pyrolysis 29
Activation ..... 31
Characterization of char and activated carbon 33
Measuring adsorption of organics from wastewater 34
7. Results and Discussion . 37
Pyrolysis of lignocellulosics and refuse-derived fuel . . .37
Activation of lignocellulosics and refuse-derived fuel ... 56
Sorptive properties of activated lignocellulosics 76
Summary 92
References 93
Appendices
A. Use of Linear Regressions for Data Plotting . . . . 101
B. Mercury Porosimetry Measurements 107
C. DOC Rate of Adsorption Experiments 115
D. Freundlich Isotherm Coefficients . 118
E. Linear Isotherms for F400 for Various Experiments 119
F. Roots of tan qn in Analytic Solution to Diffusivity 121
G. Computation of Diffusion Coefficients . . 122
Vlll
-------
FIGURES
Number
1
2
3
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
' ' Page
Schematic of pyro lysis equipment ........... ...... 30
Schematic of activation equipment ................ 32
Measured yield versus yield predicted by char yield model for
pyro lysis at 15°C/min to 500°C . , . ............... 43
Ash-free char yield versus final temperature: low and medium
heating rates ... ....... ................ 49
Ash-free char yield versus final temperature: high heating rates . 50
Carbon yield versus final temperature .............. 51
Ash-free yield versus heating rate ................ 53
Surface area of char versus final temperature .......... 57
Surface area per gram carbon versus final temperature ...... 58
Surface area per gram educt versus final temperature ....... 59
Mass loss versus activation time for prune pit char ....... 61
Nitrogen isotherms for 60M, 42M, 30M, 15M, and F400 ...... . 63
Nitrogen isotherms for F400, F100, and AN-A .......... . 64
specific surface area versus percent mass loss for
activated prune pit chars ................... 67
Mercury porosimetry: cumulative penetration volume versus pore
diameter for 60M , . .
68
Mercury porosimetry: relative cumulative pore volume of mercury
penetration versus equivalent pore radius for 60M, 42M, 30M,
15M and CHAR 69
Mercury porosimetry: relative cumulative pore volume of mercury
penetration versus equivalent pore radius for F400, F100,
and AN-A 71
Pore volume versus pore radius as determined by ^-adsorption
isotherms and mercury porosimetry for 60M, 42M, 30M, 15M,
and F400 . . . . . 72
Pore volume versus pore radius as determined by ^-adsorption
isotherms and mercury porosimetry for F400, F100, and AN-A ... 73
Kinetics of DOC adsorption for F400, FlOO, and AN-A (run 2) ... 78
IX
-------
FIGURES (continued)
Number
Page
21
22
23
24
A-l
A-2
A-3
A-4
B-l
B-2
B-3
B-4
B-5
B-6
B-7
C-l
C-2
E-l
Kinetics of DOC adsorption for 60M, 42M, 30M, and F400 (average
DOC adsorption isotherms for 60M, 42M, 30M, and F400 (run 8) ...
r& versus C£ for DOC adsorption for 60M, 42M, 30M, and F400
Pore diffusion model for kinetics of DOC adsorption by 60M,
42M 30M and F400
Ash-free char yield versus final temperature: low and medium
Ash-free char yield versus final temperature: high heating rates .
Linearity of the plots of ash-free yield versus final temperature
Linearity of the plots of ash-free yield versus final temperature
for pyrolysis at l°C/min ...............
Mercury porosimetry: cumulative penetration volume for 42M . . .
Mercury porosimetry: cumulative penetration volume for 30M . . .
Mercury porosimetry: cumulative penetration volume for 15M . . .
Mercury porosimetry: cumulative penetration volume for CHAR . . .
Mercury porosimetry: cumulative penetration volume for F400 . . .
Mercury porosimetry: cumulative penetration volume for F100 . . .
Mercury porosimetry: cumulative penetration volume for AN A . . .
Kinetics of DOC adsorption for 60M, 42M, 30M, and F400 (run 4) . .
Kinetics of DOC adsorption for 60M, 42M, 30M, and F400 (run 11) .
Linear DOC adsorption isotherms for F400 for runs 6 and 8 ....
79
81
84
89
102
103
104
105
108
109
110
111
112
113
114
116
117
120
-------
TABLES
Number
3
4
10
11
12
13
14
15
16
Page
Summary of Lignocellulosic Composition and Char Yield
Pyrolysis Conducted at 15°C/Min to 500°C 38
Summary of Results of Multiple Regressions Varying the Data
Utilized and the Inorganic Parameter in the Power Function ... 40
Composition and Char Yield of Refuse-Derived Fuel (RDF)
42
Summary of Pyrolysis of Selected Lignocellulosics for
Varying Pyrolysis Conditions . . ......... 45
Dependence of Ash Content Observed upon Ignition to
Varying Final Temperatures ...... . 47
Carbon Content and Carbon Yield for Varying Pyrolysis
Conditions 48
Linear Regressions of Ash-Free Char Yield versus Heating Rate or
ln(Heating Rate) for Pyrolysis to Final Temperature Tp . . .
Summary of Surface Area Analyses on Selected Lignocellulosic
Chars
Physical Properties of Activated Prune Pit Char and Several
Commercial Activated Carbons .....
52
55
62
Relation Between Surface Area and Burnoff for Chars Activated
for 15 Min 55
Comparison of Pore Volumes Determined by N2-Isotherms and
Mercury Porosimetry for 60M, 42M, 30M, 15M, and F400 ...... 70
Comparison of Activation of Prune Pit Char and RDF Char
Linear Adsorption Isotherm Coefficients
Estimation of the Partition Parameter R
Median Diffusion Coefficient and Pore Volume
75
82
87
90
Estimated Diffusion Coefficients ........ 91
xi
-------
TABLES (continued)
Number
A-l
A-2
D-l
F-l
G-l
G-2
Summary of Linear Regressions of Ash-Free Char Yield versus
Final Pyrolysis Temperature for Selected Lignocellulosics
Summary of Linear Regressions of Carbon Yield versus Final
f\
Computation of Diffusion Coefficients for 60M, 42M, 30M, F400
Computation for Pore Diffusion Model Based on an Assumed
Median D for 60M. 42M. 30M. F400
Page
. . 106
106
118
121
. . 122
. . 123
xii
-------
ACKNOWLEDGMENTS
The authors are grateful to Dr. Paul H. Brunner, who provided valuable
guidance on pyrolysis and on the use of gas adsorption techniques to characte-
rize porous solids. Sam Luoma, U.S. Geological Survey, Menlo Park, Calif.,
kindly permitted the use of a carbon analyzer in his laboratory. Professor
James 0. Leckie of Stanford University was instrumental in establishing the
project's direction at the outset.
xiii
-------
-------
SECTION 1
INTRODUCTION
STATEMENT OF THE PROBLEM
Solid waste management has become a national problem of prodigious pro-
portions. Historically the primary objective of solid waste management has
been disposal, while minimizing damage to the environment. However, the cost
of traditional disposal methods is now rising rapidly, while satisfactory dis-
posal sites are becoming more scarce. Thus there is widespread interest in
methods of solid waste management that will result in volume reduction, reuse
of materials, or both.
Agricultural wastes, some industrial wastes, and the major organic frac-
tion of municipal solid waste are composed of natural or modified plant tissue
(lignocellulose). Awareness of the large potential of such materials as a
source of energy and raw materials has aroused interest in the physical and
chemical processing of lignocellulosic wastes. Pyrolytic processes, in par-
ticular, appear to offer the combined benefits of reducing the solid waste
volume, minimizing pollutant emissions, and producing valuable products such
as gaseous and liquid fuels and a solid, carbonaceous char.
Recently such pyrolytic chars have been considered as possible source
materials for production of adsorbents for water and wastewater treatment.
The economic" viability of powdered activated carbon processes for municipal
wastewater treatment may depend on the availability of an,inexpensive, one-use
adsorbent. Additionally, the proposed Environmental Protection Agency regula-
tions for water quality control regarding organic pollutants encourage the use
of adsorption processes to meet these standards, thereby creating a large need
for activated carbon or similar adsorbents. Conceivably, this potential
market could be supplied by the conversion of abundant lignocellulosic wastes
into inexpensive activated carbon with suitable adsorptive properties.
OBJECTIVES
The objectives of this study were to:
1. Determine the effect of lignocellulosic composition and pyrolysis con-
ditions on the yield and surface physical properties of pyrolytic
char.
2. Ascertain the yield and properties of activated carbon prepared by
activation of a lignocellulosic char.
-------
3. Determine the usefulness of data obtained from experiments with ligno-
cellulosic substances for understanding the pyrolysis and activation
of municipal solid waste.
4. Compare activated carbons prepared from lignocellulosic waste with
commercially available activated carbons according to surface physical
characteristics (e.g. surface area), sorptive capacity, and the kinet-
ics of the uptake of organic carbon from secondary effluent.
RESEARCH APPROACH
The study of pyrolysis was conducted with a wide range of materials rep-
resentative of solid wastes of agricultural, forest product, industrial, and
municipal waste origin. The activation studies were performed on a char pre-
pared from a specific lignocellulosic material (prune pits), as well as class-
ified municipal solid waste. Pyrolysis and activation were performed.in
small-scale laboratory equipment to ensure controllable, well-defined condi-
tions .
The adsorption comparisons of waste-derived and commercial activated
carbons were conducted in small-scale laboratory apparatus using filter-
sterilized, unchlorinated secondary effluent from a local municipal wastewater
treatment facility. Comparisons were made on the basis of adsorption equilib-
rium capacity for and kinetics of removal of dissolved organic carbon (DOC).
Results of these adsorption experiments were interpreted in terms of simple
models for adsorption and transport.
_
-------
SECTION 2
CONCLUSIONS
The yield of char from pyrolysis of lignocellulosic materials (agricultu-
ral wastes, paper, and modified cellulose) can be correlated with the initial
composition. Data analysis by multiple linear regression reveals that the
yield,can be expressed as a linear combination of the contributions from the
holocellulose, lignin, extractive, and ash fractions, with a correlation
coefficient exceeding 0.99.
The char yield (mass basis) from lignin (50 to 55%) is higher than that
from the other organic fractions: holocellulose (19%) and extractives (35 to
45%). The ash fraction appears to behave as an inert material during pyroly-
sis. No significant catalytic effect due to ash constituents was observed in
pyrolysis of natural lignocellulosic materials.
The measured char yield from refuse-derived fuel (RDF, an air-classified,
shredded municipal solid waste fraction) was 10% lower than the value predic-
ted for its composition, based on the yield correlation for lignocellulosic
materials (34.2% compared to a predicted value of 37.5% under the conditions
studied). Adjustment for the plastic content of RDF reduced the difference
between the measured and predicted values to less than 4%.
The char yield decreased approximately in linear dependence on the pyrol-
ysis temperature (above 500°C) and on the logarithm of the heating rate.
The specific surface areas (CO^-BET, 195K) of chars from lignocellulosic
materials ranged from 300 to 650 m per g char. The specific surface of chars
prepared from representative plant wastes, a paper product, and a purified
cellulose were similar in magnitude and showed a similar dependence on pyroly-
sis temperature. After adjustment for carbon content, the values of specific
surface (m per gram carbon) for five lignocellulose-derived chars agreed
within ± 15%. The specific surface of the char increased markedly (20 to
30%), when the pyrolysis temperature was raised from 500 to 700°C, but gener-
ally was not significantly higher at 900°C than at 700°C.
i
Activated carbon prepared by COo activation of a char obtained from a
representative lignocellulosic material (prune pits) had a specific surface
area in the range commonly found for commercially available activated carbon
products. The specific surface of activated carbon derived from prune pits
was 660 to 1700 m2/g (N2-BET, 77K) depending on the time of activation at
900°C (15 to 60 min).
-------
Activation of the prune pit char for 42 to 60 min at 900°C resulted in an
activated carbon with an ^-BET surface area greater than that of ground
Filtrasorb 400, an activated carbon widely used for treatment of water and
wastewater. The volumes and size distributions of pores in the range of 3 to
300 nm were virtually indistinguishable, when a highly activated carbon of
lignocellulose origin (60 min activation time) was compared with the ground
commercial product, Filtrasorb 400. Activation of prune pit char for 30 min
at 900°C resulted in a product with surface area and pore size characteristics
similar to Aqua-Nuchar A, a low-cost powdered activated carbon used in water
treatment, but inferior to Filtrasorb 400.
Activated carbon prepared from refuse-derived fuel (RDF) had a specific
surface less than half that of activated carbons derived from prune pits under
similar conditions. This is largely explained by the high inorganic content
of RDF.
Activated carbons derived from prune pits adsorbed substantial quantities
of organic constituents (measured as DOC) from secondary effluent. The equi-
librium adsorption capacity of prunepit activated carbon depended strongly on
the activation time; the DOC uptake capacity after 60 min activation was
approximately 2.5 times greater than that after 42 min, and five times greater
than after 30 min. The equilibrium capacity of Filtrasorb 400 was intermedi-
ate between those of prune pit carbons that had been activated for 42 and 60
min.
The adsorption equilibrium isotherms for DOC uptake from secondary efflu-
ent were approximately linear, after adjustment was made for a non-adsorbable
fraction. The resulting partition coefficients lie in the range of 20,000 to
100,000 g DOC adsorbed per g DOC in solution within the adsorbent grains.
The apparent pore diffusion coefficients estimated for the prune-pit
activated carbons were in the range of 1.1 x 10 m Is (30 min activation) to
3.2 x 10 m /s (60 min activation). The corresponding value for Filtrasorb
4001.4 x 10 m /sfell within that range. The diffusion coefficients
estimated from experimental data agree in order of magnitude with expectations
based on molecular diffusivities. Accordingly, a simple pore diffusion model
appears to explain the observed DOC uptake rates.
Powdered activated carbon prepared from lignocellulosic waste material
(prune pits) compares favorably with commercially available activated carbon
products as an adsorbent for removal of dissolved organic carbon from second-
ary effluent. In view of the similar properties and yields of chars prepared
from a broad spectrum of lignocellulosic materials, it is possible that adsor-
bents useful for water and wastewater treatment could be produced from any of
a wide variety of lignocellulosic solid wastes from agriculture and the .forest
products industry. Municipal solid waste appears less suitable as a raw
material for activated carbon manufacture than are agricultural and.wood
wastes.
-------
SECTION 3
RECOMMENDATIONS
In this work it has been shown that activated carbons prepared from an
agricultural waste material can perform as effectively in removing organic
constituents from wastewater effluent as do presently used, commercial acti-
vated carbon products. Hence, it seems justified to conduct further research
and development directed toward: identifying and characterizing candidate raw
materials, evaluating their availability, optimizing the conditions of pyroly-
sis and activation, understanding the effects of physical and surface chemical
characteristics on adsorbent performance, verifying the validity of laboratory
tests of adsorbent performance, and,comparing the efficiency of waste-derived
and conventional activated carbons for the removal of organic priority .pollu-
tants from water and wastewater.
Agricultural wastes constitute a solid waste management problem, but also
a resource of lignocelluloslc materials of high carbon content. If agricul-
tural products such as grain are to be converted to liquid fuels using fermen-
tation, large quantities of lignocellulosic wastes will result. The utiliza-
tion of these solid wastes as pyrolysis feedstocks ought to be investigated.
Activated carbon production is one promising pyrolysis alternative. The
logistics and economics of such operations deserve evaluation.
, Additional work is required to optimize the overall sequence of pyrolysis
and activation. The approach in this study has been to optimize pyrolysis
with the objective of maximizing the micropore volume of the char, while
maintaining an acceptable char yield. In activation, only one activating
atmosphere and one temperature were used. A-parametric study of activation
temperature and atmosphere may result in significant improvement in the yield
and properties of activated carbon from solid wastes. Also, it is possible
that the overall yield and the product properties could be improved by pyro-
lyzing under conditions less severe than needed to prepare a char having a
maximum micropore volume. This approach merits further research.
The relation between an adsorbent's physical and chemical properties and
its adsorption performance must be better understood if adsorbents prepared
from a great number of candidate solid waste raw materials are to be screened
intelligently. Basic questions relating to the interpretation of data from
gas penetration and mercury porosimetry measurements need to be clarified
before such information can be used with confidence.
.From this work, it appears that a simple approach based on linear adsorp-
tion equilibrium, coupled with a transport model incorporating pore diffusion
in spherical geometry, is sufficient to simulate the extent and rate of uptake
-------
of dissolved organic carbon (DOC) from secondary effluent, if biodegradation
is excluded. Because of the convenience afforded by this simple model, its
general applicability to water and wastewater treatment should be investi-
gated. If verified, the model would facilitate the computation of DOC removal
in treatment processes.
In the light of increased concern about hazardous organic contaminants,
solid-waste-derived activated carbons should be tested to ascertain their
efficacy for adsorbing selected organic priority pollutants. Also, the leach-
ing of priority pollutants (both inorganic and organic) from waste-derived
activated carbons should be compared to that from currently marketed products.
-------
SECTION 4
PHYSICAL CHARACTERIZATION AND PREPARATION OF ACTIVATED CARBON
Activated carbon is generally produced by a two-step process consisting
of pyrolyzing (carbonizing) the source material under appropriate conditions
and then activating the resultant pyrolytic char by oxidation in a controlled
environment (1). The purpose of these processesHis to yield a final product
with extensive porosity and high adsorptive capacity. In this chapter we
discuss pyrolysis, activation, and methods of characterization of the extent
of porosity. Physical characterization will be discussed first to allow
definition of terms used in subsequent sections.
PHYSICAL CHARACTERIZATION
The pores in chars and activated carbons vary in size from remnants of
the tissue structure in the case of lignocellulosics (cell diameters range
from 10-100 ym) to apertures which are inaccessible even to helium at room
temperature (2,3,4). The volume which constitutes the internal pores of an
adsorbent and is accessible from the exterior of the adsorbent particle is of
prime interest in studies of sorptive capacity and behavior. A large pore
volume does not necessarily imply a large pore surface area, because the
specific surface area is dependent on the distribution of pore sizes (4); a
preponderance of small pores is necessary to assure a large specific surface
area. For porous adsorbents, the following classification of pore sizes is
common (4,5):
Micropores diameter < 2nm
Transitional-Pores 2 nm < diameter < 20nm
Macropores 20 nm < diameter
The methods used in this study to characterize the internal pore structure of
chars and activated carbons are based on adsorption of gases and mercury
penetration, discussed in detail below.
Gas Adsorption
Gas adsorptipn is a common technique for the estimation of surface area
of porous materials and can also be used to estimate pore size distribution,
at least for micro- and some transitional-porosity.
The gas adsorption analysis entails admitting the adsorbate gas to a
sample of known weight, which.has previously been completely freed of all
adsorbed gases and vapors during an outgassing period (elevated temperature
and high vacuum). In practice the gas is admitted in known incremental
amounts, determined by allowing the gas to come to equilibrium in the
-------
instrument manifold (known volume and temperature) and measuring the pressure.
The number of moles of gas present is estimated by means of the ideal gas
law. The gas is then allowed to expand into the previously evacuated sample
container (the volume and temperature of the sample container are known).
Some portion of the gas adsorbs onto the sample and eventually a new equilib-
rium pressure is reached. A material balance on the adsorbate using the gas
law, with corrections for nonideality due to low sample temperature, allows
the determination by difference of the number of moles of gas adsorbed by the
solid sample. The procedure is repeated for increasing equilibrium pressures,
generating the adsorption isotherm, or the amount of gas adsorbed (expressed
as cm of gas (STP) per gram sample) as a function of equilibrium pressure.
The resultant adsorption isotherm must then be interpreted to yield
estimates of sample surface area, pore size, and pore volume. The interpre-
tive methods in common use are discussed separately below, followed by a
discussion of two very important details of the analytical procedure: outgas-
sing conditions and equilibrium time.
BET Surface Area
This method is applicable to adsorption data from nitrogen at 77K and
carbon dioxide at 195K. The method assumes layer-by-layer filling of the
pores with the adsorbate gas. The sample surface area is calculated by deter-
mining the number of molecules of adsorbate necessary to produce a monomolecu-
lar layer on the sample, and multiplying that number by the cross sectional
area assumed to be occupied by the adsorbed gas. The monolayer capacity is
conventionally determined by the Brunauer-Emmett-Teller (BET) equation (4,6),
used in the following form:
(1)
V(p - p) V C V C p
s mm s
where V = the volume adsorbed at equilibrium pressure p,
Ps m the saturation pressure of the adsorbate at the adsorption tempera-
ture,
Vm = the monolayer capacity, and
C -a constant.
A plot of p/V(pg - p) versus p/p therefore should yield a straight line of
slope (C - l)/VmC and intercept I/VmC. The range of relative pressure for
which the plot is linear will vary with the material being analyzed (4); in
general the linear range is (0.05 < p/p < 0.35) (7).
S
The monolayer capacity is thus:
V
m
slope + intercept
(2)
-------
Sample surface area is therefore given by the following expression
S =
V NA a
m A a
(3)
V
where S = sample surface area (m /g),
V = monolayer volume at STP (cm /g sample),
m
o
a = cross sectional area of adsorbed molecule (m /molecule),
3. "'
*3
V° = ideal gas molar volume at STP (cm /mole), and
Ni
Avogadro's number (molecules/mole).
The BET method was originally derived for non-porous or large-pore adsor-
bents. The method assumes that the surface is covered with energetically
uniform sites, that only one molecule of adsorbate is adsorbed to each site,
and that there is no interaction between adsorbed molecules (7). There is
general agreement that these assumptions may not be valid for some adsorbate-
adsorbent systems (4,5,7). In particular, the assumption of layer-by-layer
adsorption is thought to, be inapplicable to microporous adsorbents, such, as
pyrolytic chars and activated carbons, because the diameters of the pores
(< 2 nm) are in the same;order of magnitude as those of the adsorbing gas
(e.g., about 0.5 nm for ^ at 77K) (5,7,8). Dubinin (5) states that due to an
adsorption force field in the entire volume of micropores, adsorption results
in volume filling in the pores. Since this results in larger volumes of
adsorbed gas for a given pressure than would be the case for monolayer adsorp-
tion, the BET method will calculate erroneously high values of "monolayer
volume" and "surface area" (4,5,9,10). The theoretical maximum surface area
of pure graphite is 2680 m /g, assuming adsorption on both sides of the gra-
phite plane (9). Thus, surface areas calculated by the BET method which are
greater than 1000-1300 m /g must be erroneous (9,11).
The reporting of surface areas for microporous adsorbents is therefore
open to criticism. The situation is compounded by the ambiguity inherent in
assigning a value for the cross-sectional area of the adsorbed molecule (4,7).
Due to geometric and energetic characteristics which are unique to each adsor-
bate-adsorbent system, the monolayer packing density and thus the effective
cross-sectional area of an adsorbate may vary with the adsorbent (7). Since
it is impossible to allow for these considerations in general, it is common to
assign cross-sectional areas of adsorbates based on comparative adsorption
experiments on well-characterized adsorbents; the adsorbate generally used as
the standard is N2 at 77K (7).
In summary, there are substantial uncertainties in the calculation of BET
surface area for microporous materials such as pyrolytic chars and activated ,
carbons. Moreover, the very concept of a "surface" is questionable for pores
whose dimensions in such materials, often < 0.6 nm (8), approach the carbon-
carbon bond lengths (~ 0.14 nm) in a graphitic structure. Nonetheless, pro-
vided the ambiguity of the concept and calculation technique is recognized,
-------
BET surface areas are useful comparative measures of porosity in chars and
activated carbons, and are commonly reported for lack of a more accurate
method of characterization.
Pore Size Distribution
The pore radius (r ) corresponding to the given point on the adsorption
isotherm (i.e.,- a given value of relative pressure) may be calculated from the
modified Kelvin equation (4):
r = r, + t
P k'
(2x10 )q V cos 9
RT An(p /p)
s
+ t
(4)
where r-
k
a
V
0
PS
P
R
Kelvin radius (nm),
surface tension of the liquid adsorbate (N/m),
o
molar volume of the liquid adsorbate (m /mole),
contact angle between the liquid and pore wall,
saturation vapor pressure (Pa),
equilibrium pressure (Pa),
gas constant (J/K*mole),
T - absolute temperature (K), and
t - layer thickness, discussed below, (nm).
It is commonly assumed that © equals zero, i.e., that the liquid adsor-
bate wets the pore walls (4). Use of the Kelvin equation assumes filling of
the pores by capillary condensation. Before capillary condensation occurs,
however, one or more adsorbed layers may form on the pore walls (at least for
pores > 2 nm diameter). Thus the actual pore radius (r ) will be the Kelvin
radius plus the thickness of the already adsorbed layers. Various methods
have been used to estimate the layer thickness; in this study we used the
method of Cranston and Inkley (12).
For nitrogen adsorption at 77K, equation (4) reduces to
r =
0.96
/p)
+ t
.(5)
The range of pore sizes that can be determined with reasonable accuracy
by this approach is approximately 2 nm < diameter < 50 nm (7). The upper
limit is determined by the shape of the isotherm near saturation and the
precision of gas measurement and temperature control. The lower limit is due
to the inappropriateness of assuming the existence of a liquid meniscus and
10
_
-------
bulk liquid properties for capillary "condensates" in pores whose diameters
correspond to a few adsorbate molecular diameters.
There is some controversy regarding whether to use the adsorption or the
desorption branch of the isotherm to calculate pore size. Although use of the
desorption branch is customary, for materials with "ink bottle" type pores
(narrow neck, larger pore beyond), analysis of the adsorption branch yields a
more nearly correct interpretation of the porosity (7). Many chars are
thought to contain such "inkbottle" pores (13).
Pore Volume '
Combination of the basic adsorption data with the Kelvin equation theo-
retically can be.used to estimate the pore volume for pores smaller than a
given diameter or to calculate the complete pore volume distribution (14).
Cranston and Inkley (12) describe a method for the determination of pore
volume distribution which takes into' account the fact that the volume deter-
mined in the adsorption experiment includes both that which fills the pores
with radius less than r plus that which is adsorbed on the walls of pores
with radius greater than r . The raw adsorption data, however, are expressed
as volume (at STP) of adsorbate gas adsorbed in the pores. To convert this to
an estimate of the absolute volume of the pores, the density of the adsorbed
gas must be known. As discussed previously, this may depend on the particular
adsorbate-adsorbent system. Although there is general agreement on the use of
bulk liquid density for N^ at 77K, the appropriate density for CC>2 at 195K is
still an unresolved question (9).
Outgassing Conditions
Gas adsorption data obtained from microporous chars and activated carbons
are very sensitive to the heat treatment and outgassing conditions used to
prepare the material for analysis (4). The materials are thought to chemisorb
oxygen on exposure to air at room temperature, e.g., during storage. Since
chars contain "ink bottle" pores, the chemisorbed oxygen may reduce the size
of the "bottleneck" to such an extent that the adsorbate cannot enter, result-
ing in lower surface area and pore volume values. It has been shown (4) that
apparent surface area increases as outgassing conditions become more severe
(higher temperature or longer evacuation). Thus, to ensure absolute and re-
peatable data, very severe conditions are recommended, such as evacuation at a
temperature close to but lower than the temperature at which the char was pre-
pared (4). Unfortunately, such outgassing may, itself, alter the char pore
structure (15).
Equilibration Time
In the analysis of microporous adsorbents, the time to reach adsorption
equilibrium can be quite long due to activated diffusion, i.e., diffusion of
the adsorbate through pores only a few times the adsorbate's diameter (4, 16).
For this reason the socalled equilibrium pressure is generally noted after a
fixed arbitrary time, depending on the adsorbate, or noted when the rate of
decrease of pressure reaches some fixed level. If equilibrium times are very
11
-------
long there is a danger of error caused by the leak rate characteristic of the
analytical apparatus.
Comparison of Adsorbates and Adsorption Temperatures
Due to the size of the nitrogen molecule, some microporosity, especially
in chars, may not be accessible to it during the adsorption analysis (4).
Accordingly, the estimate of surface area based on the adsorption of N£ at 77K
is considered to exclude the area of the smaller micropores, diameter approxi-
mately 0.6 nm or less (4). Adsorption of CC>2 at 195K is thought to measure
essentially the total surface of microporous chars and carbons (4).
Mercury Porosimetry
Mercury porosimetry is used to measure the macro- and transitional-pore
volume distribution of porous materials including carbonaceous adsorbents
(4, 17). In practice mercury is forced into the pores of the material at
increasingly higher pressures. The volume of mercury penetration is measured
as a function of the applied pressure (p). The value of the pore radius (r)
corresponding to (p) is calculated using the Washburn equation (18):
-(2y cos
(6)
where Y is the surface tension of mercury and 0 is the contact angle between
mercury and the pore wall.
The analysis is conducted by adding a known weight of sample to the
penetrometer followed by sample evacuation. Next mercury is admitted to the
evacuated chamber and the pressure-penetration measurements are made. Pres-
sures as high as 420 MPa (60,000 psi) are possible with commercially available
equipment, which theoretically implies an ability to measure pore sizes down
to 1.8 nm. However, Mahajan and Walker (4) state that pore size information
for high pressures may be faulty due to particle breakdown and/or opening of
previously closed pores. Dickinson (19) reports that mercury intrusion can
cause such damage to graphite for pressures above 2000-3000 psi, corresponding
to 50-35 nm pore radii.
The Washburn equation assumes cylindrical pores. Chars, however, are
known to have "ink-bottle" (or aperture-cavity) type porosity (4); for such
pores, the pore volume distribution results will be subject to error because
for a given pressure the pore radius will be determined by the aperture, while
the intruded volume will be determined by the cavity.
Other limitations are discussed by Mahajan and Walker (4), including the
possible dependence of surface tension on pore radius for radii < 50 nm.
Finally, for powdered samples mercury porosimetry results cannot distin-
guish between the small interparticle voids and macroporosity in the particles
themselves. In this case the data must be analyzed carefully to avoid misin-
terpretation.
12
-------
Thus, as was the case for the other methods of physical characterization
of chars and activated carbons, interpretation of mercury porosimetry results
is subject to a number of limitations. However, if the limitations are ack-
nowledged, the technique is useful for the comparison of materials.
PYROLYSIS
Pyrolysis is strictly defined as chemical decomposition by heat. As a
step in activated carbon preparation, pyrplysis is generally conducted at
temperatures in the range 400°-1000°C and in the absence of air (1). Research
on pyrolysis of carbonaceous materials, especially with respect to char prop-
erties, has been largely limited to coal and model compounds such as cellu-
lose. The focus of this study is utilization of cellulosic and ligocellulo-
sics waste materials. Pertinent concepts of pyrolysis of pure cellulose and
lignocellulosics in general are discussed below.
Pyrolysis of Cellulose
Cellulose is the main structural component in the cell walls and fibrous
and woody tissues of plants. It is a linear polymer of D-^glucose in gl -> 4
linkage, usually represented by the chemical formulas (CgH^Q^S^n* Naturally
occurring cellulose polymers contain 300 to 15,000 glucose monomer units and
are organized in bundles of parallel chains to form fibrils (20).
Despite the fact that the pyrolysis of cellulose is a well-examined
process (21), no definite scheme of reaction mechanism has been established.
This is due to the highly complex nature of the thermal decomposition of
cellulose, which consists of many interrelated reactions with a great number
of reactants, intermediates, and reaction products* Tang and Bacon (22,23)
have presented a model for the pyrolysis of cellulosic fibers, which is in
agreement with most of the work done by others (21). Based on carbonization
experiments and subsequent elemental analysis, x-ray, infrared, and thermal
analyses, they conclude that pyrolysis starts with the physical desorption of
water« 150°C), followed by dehydration (150°-240°C) and cleavage of the 1 -»
4 glycosidic linkages (> 240°C). According to Tang and Bacon, the final poly-
meric char is made up of four-carbon building blocks, which orginate directly
from the initial cellulose. Thus the maximum char yield is expected to be
29.5%. The .dehydration reactions below 240°C appear to be slow when compared
to the depolymerizatlon and scission of C-0 and C-C bonds above 240°C.
f
The reaction kinetics of the thermal decomposition of cellulose has been
widely studied in the context of fire research and flame retardants (21,24,
25). Tang and Neill (25) found that the pyrolysis reactions are best separa-
ted into two groups: a pseudo-zero-order reaction below 310°C with an activa-
tion energy of - 34 kcal/mol, and a pseudo-first-order reaction above 310°C
with a higher activation energy of 54 kcal/mol. Flame retardants and some
inorganic contaminants (such as Lewis acids) have been shown to decrease the
activation energies as well as the DTA maxima and usually to increase char
yield (21,25-28).
13
-------
The yield and properties of low-temperature (< 1000°C) pyrolytic chars
have not been as well examined as the pyrolysis reaction itself. In a study
of cellulose pyrolysis and activation (26), Brunner found a log-linear depen-
dence of char yield on the rate of heating to the final pyrolysis temperature.
Slow heating (e.g., < l°C/min) was found to result in considerably increased
char yields. This was thought to result from the longer exposure of the cel-
lulose to temperatures below 240°C, in which range the dehydration reactions
predominate. The more completely dehydrated polymer, due to the presence of
carbon-carbon double bonds, is thought to be less susceptible to cleavage at
temperatures above 240°C, resulting in the observed higher char yields.
It has also been shown (29) that, when cellulose was pyrolyzed at a slow
heating rate, there was not significant weight loss above approximately 700°C,
whereas rapidly charred cellulose continued to lose weight at least up to
1000°C. Furthermore, slow heating resulted in lower oxygen content of the
char.
The surface area and micropore volume of cellulose char is reported to
increase with temperature up to a maximum (700°-1000°C), beyond which they
begin to decrease (26,29,30). This effect, also observed for pyrolysis of
coal products, is thought to be due to progressive development of micropores
up to 1000°C followed by rapid closure of the micropore entrances (4,11,
30,31). Constriction of the micropores is detectable by gas adsorption meth-
ods as low as 700°-800°C (11,29). Furthermore, it has been observed that low
heating rates result in a higher maximum surface area, somewhat larger micro-
pore openings, and a less pronounced decrease of surface area at higher tem-
peratures (26,29).
Finally, pore development in pyrolyzed cellulose is limited to small
raicropores whose diameters are on the order of magnitude of the molecular
sizes of nitrogen and carbon dioxide (29), approximately 0.4 nm or less
(7,16). Mercury porosimetry of cellulose chars heated at various rates to a
range of final temperatures (500°-1000°C) revealed no development of transi-
tional- or macroporosity (29).
Pyrolysis of Lignocellulosics.
Plant tissue, or lignocellulose, is a matrix of three.main components:
lignin, cellulose and hemicellulose. Cellulose, as mentioned previously, is
present in long fibrils, while the other two components fill the interfibril-
lar spaces and serve to cement the matrix together (20). Lignin is a high
molecular weight (10 10 ) threedimensional polymer of aromatic alcohols
(32). Hemicellulose refers to branched polysaccharides composed primarily of
pentoses with lesser amounts of hexoses (20). Both the hemicellulose and
lignin fractions vary chemically with the parent lignocellulosic and are
difficult or virtually impossible to isolate unaltered (20,32). Compared to
the substantial body of research on cellulose pyrolysis, considerably less is
known about the pyrolysis of the other two components and very little about
pyrolysis of the lignocellulosic matrix in general.
Shafizadeh and McGinnis (33) found that the thermal behavior of wood
reflects the sum of the thermal responses of its three major components,
14
-------
cellulose, hemicellulose, and lignin. The data indicate, as discussed by
Shafizadeh and Chin (34), that the components are initially dried on heating
at 50°-100°C. Hemicellulose is the least stable component, decomposing at
225°-325°C. Lignin decomposes gradually within the wide range of 250°-500°C,
with a much higher percent char yield than the other components. They con-
clude further that since the pyrolysis products of wood reflect the sum of the
products from the major components, there is no major interaction between the
components during pyrolysis.
Philpot (35), however, in a study of a variety of plant materials, found
that minerals present in the plant tissue appeared to increase the yield of
char from the lignocellulose, much as fire retardants and certain inorganics
had been shown to affect cellulose pyrolysis (21,25-28). He found, further,
that the char yield was even better correlated to silica-free ash, a measure,
of the mineral content of the plant tissue excluding silica. The explanation
offered for this correlation was that silica, unlike the other ash components,
is effectively inert and incapable of affecting pyrolysis of the organic
fraction. ,
Philpot and others (35,36) have shown for pyrolysis of plant materials
that the effect of inorganics on char yield levels off for ash contents be-
tween 5-7%. Also it has been observed (34) that the effect of inorganics on
hemicellulose pyrolysis is similar to the effect on cellulose. Little is
known about the effect of inorganics on char yield from lignin, although
strong acid treatment of lignin is known to increase its char yield slightly
(32,33).
Rothermel (37), using data derived from other studies (36,38), developed
an empirical model for lignocellulosic char yield (pyrolysis at 15°C/min to
4QO°C) as a function of composition. The model calculated the total char
yield as the sum of the char yields of the components, and assumed the compo-
nents did not interact, except that the holocellulose (cellulose plus hemicel-
lulose) yield increased with silica-free ash content. The catalytic effect of
the silica-free ash was included as a power function to account for leveling
off of the effect at increasing ash contents.
There has been little reported research on surface area and pore volume
development in lignocellulosic chars, only studies specific to one material
and set of pyrolysis conditions (e.g., olive stones (2), and plum stones
(3)). Based on these limited studies, it appears that lignocellulosic chars
develop only microporosity.
There has been no systematic research reported on the effect of pyrolysis
heating rate or final temperature on the yield, surface area or micropore
volume of lignocellulosics in general.
ACTIVATION
Activation refers to processes which increase the adsorptive capacity of
chars, usually by increasing the extent of porosity (surface area and micro-
pore volume). Many commercial processes entail reacting the char with
15
-------
oxidizing gases (e.g., steam, carbon dioxide, oxygen or air) at elevated
temperatures (600-1000°C)(1). In general, the adsorptive capacity developed
is determined by the nature and concentration of the oxidizing gas, the tem-
perature of the reaction, the extent of oxidation (measured for example as
weight loss, or "burn-off"), and the amount and kind of mineral ingredients in
the char (1). It is also generally accepted (1,39,40) that the results of
activation depend on the nature of the starting char, its processing history
and to some degree its parent material.
There does appear to be some agreement (4,41-44) that gasification (acti-
vation by oxidizing gases) may involve any or all of three-basic phenomena:
widening of porosity existing in the char, opening of previously blocked
micropores and, possibly, creation of or elongation of micropores. Thus, as
activation proceeds, both the total number of pores and their average radius
is increased, resulting in an increase in specific pore volume and specific
surface area (i.e. per gram remaining material) (4). At some point, however,
depending on the structure of the char, walls between existing pores are
gasified away, resulting in a decrease in the total number of open pores
(4,43). While this leads to a continuous increase in specific pore volume,
specific surface eventually reaches a maximum and declines thereafter (4,45).
One study, (46) for example, reports such maxima in the range 60-80% weight
loss in activation for coal chars. Brunner (26), however, in a study of
cellulose char activation, found a continuous increase of surface area with
increasing oxidation up to weight losses of 60-80%.
The choice of oxidizing gas also affects the porosity developed during
activation. Carbon dioxide activation has been observed to produce only
micropores for burn-offs less than 30-35% in studies of cellulose, cellulose
triacetate and sucrose chars (26,41,47). Steam activation has been found to
result in more pronounced transitional-porosity (42,48). Tomkow, et al. (48)
compared Oo, C02 and t^O activation of brown coal chars and found the effect
of the gas varied with the burnoff. At low-burnoff (1-8%), oxygen activation
resulted in higher surface areas than carbon dioxide or steam. At high burnoff
(70%), the reverse was found. For oxygen activation, total pore volume lev-
eled off at only 25% burnoff, whereas total pore volume increased steadily for
steam and carbon dioxide activation. Carbon dioxide activation yielded the
highest total pore volume at high burnoff.
Both the partial pressure and flowrate of the oxidizing gas has been
observed to influence the rate and effect of gasification. Carbon monoxide, a
product of carbon dioxide activation, is known to have a pronounced retarding
effect on the gasification rate (49,50,51). At low flow rates of C02, the CO
retained at the carbon surface is thought to inhibit gasification of the
particle exterior and result in greater development of microporosity (50).
Also, dilution of carbon dioxide with inert gases (e.g. Xe, N2, Ar) has been
observed to result in higher gasification rates (52).
Activation temperature also profoundly affects the rate of gasification.
Brunner (26) found that the time for 40% burnoff of cellulose char in C02
increased twofold and threefold as temperature was lowered from 960°C to 915°C
and 880°C, respectively. Similarly, the rate of development of surface area
increased with activation temperature.
16
-------
Finally, the rate of gasification is known to increase in the presence of
almost any inorganic impurity, even in trace amounts (51). It is thought that
inorganic impurities in the char agglomerate as reduced metals during activa-
tion (53) and migrate to the char surface (51). Catalysis occurs only in the
vicinity of the metal agglomerate (53,54) and results in development of macro-
and transitional-pores with little change in microporosity (53). Catalysis is
thought to continue until the metal is oxidized (53). However, the same
activation products that normally inhibit uncatalyzed gasification (i.e., CO,
H2) may serve as accelerators in that their presence in sufficient amounts
maintains the catalyst in a more reduced and active state.
17
-------
SECTION 5
MEASURES OF SORPTION PERFORMANCE
Sorption performance measurements include kinetic studies and equilibrium
isotherms. Kinetic studies relate to the rate at which solute diffuses into
activated carbon grains and is adsorbed onto the surface of the activated
carbon; equilibrium isotherms measure the partitioning of solute between that
which adsorbs and that which remains in solution at equilibrium conditions.
EQUILIBRIUM ISOTHERMS
Several models have been proposed to describe adsorption isotherms: the
Langmuir, Brunauer-Emmett-Teller (BET), Freundlich, Linear, and Simplified
Ideal Adsorbed Solution (IAS) models.
The Langmuir isotherm (55) was derived in much the same way that chemical
equilibrium equations have been derived, assuming a four-component system of
adsorbed solvent, solute in solution, adsorbed solute, and solvent in solution
(56). The Langmuir isotherm is of the form:
qe ~
1 + bC
(8)
where q is grams of solute adsorbed per gram of activated carbon, and Q is
the grams of solute adsorbed per gram of carbon at complete monolayer coverage
of the carbon. C is the concentration of solute in bulk solution, and b is a
constant related to the net enthalpy, AH, of adsorption.
The Langmuir model occasionally has been used to describe adsorption onto
activated carbon from a-liquid phase with some success (57,58). The Langmuir
equation is based on the following assumptions: The maximum adsorption that
can ever occur corresponds to a saturated monolayer of solute molecules on the
adsorbent surface; the energy of adsorption is constant; and no surface diffu-
sion occurs (59). In general, those conditions are not fulfilled in most
solute/activated carbon systems. Since adsorption forces pervade the micro-
pores of carbon (5), volume filling may occur instead of monolayer coverage.
Secondly, the surface of activated carbon is composed of many types of func-
tional groups (60), which exhibit a broad spectrum of adsorption energies.
Thirdly, surface diffusion is believed to occur (13,61).
The BET model is an extension of the Langmuir model. It is based on the
assumption that a number of layers of adsorbate form on the surface of a solid
18
-------
*and that for each of these layers, the La'ngmuir equation applies. The BET
model is used to describe the behavior of gases adsorbing onto a solid when
the gas is at temperatures close to those required for condensation of the
gas. Section 4 of this report discusses the BET model in more detail.
The Freundlich isotherm (62) makes allowances for heterogeneous surface
energies and is of the form
Kf Ce
1/n
(9)
where Kf is a constant related to sorption capacity, and 1/n is related to the
favorability of adsorption. If n > 1, favorable adsorption is indicated. If
n < 1, adsorption is unfavorable. If n = 1, the isotherm is linear, i.e., the
amount of solute adsorbed is directly proportional to the amount of solute
present.
The Freundlich isotherm has been used extensively in both a theoretical
and empirical context, and often successfully describes the adsorption behav-
ior of a wide range of organic compounds from solution phase onto activated
carbon. Sontheimer (63) has found the Freundlich isotherms more useful than
Langmuir relations to characterize the adsorption of a wide range of com-
pounds; Dobbs et al. (64) have presented a collection of Freundlich isotherms
for 143 organic compounds.
The linear isotherm is of the form:
It is a special case of the Langmuir equation (equation 8) where
, Q°b = R (10)
1^ ,
and bCe « 1, or of the Freundlich equation (equation 9) where 1/n = 1.
The ideal adsorbed solution model (IAS) was developed for bi-solute
'systems by Radke and Prausnitz (65). This was later applied and simplified by
Fritz and Schliinder~(66) . Radke and Prausnitz (65) proposed from thermody-
namic principles that the adsorbed phase could be considered as an ideal two-
dimensional solution; it could therefore be described by equations in two
dimensions similar to those for bulk solution in three dimensions. Further,
if it is assumed that the spreading pressure, u, of each of several solutes in
mono-solute experiments is equal to the spreading pressure of the mixtures of
these solutes in bi-solute or multi-solute experiments, then the data from the
former can be used to predict the adsorption behavior of various compounds in
multi-solute experiments. The spreading pressure IT is an integral function of
bulk and solid concentrations; in principle, it can be determined for each
component independently of the other components by mono-solute isotherms.
19
-------
Fritz and Schlunder (66) simplified the IAS model by assuming that the
adsorption of each individual component could be modeled by the Freundlich
isotherm. This simplified the integral for TT found in Radke and Prausnitz ,to
a more useful form:
Xl =
- C
32,k
2,k
(Ha)
(lib)
X2 = X2°(l -
(lie)
= 1
(Ud)
where Yi° and
are adsorbed concentrations of component 1 and 2, respec-
tively, determined in independently determined mono-solute experiments; X-^ and
Y-j are the equilibrium bulk and adsorbed concentrations of component 1 in the
bi solute system and b and C are constants as described in (66). The value z-^
is the mole fraction of component 1, excluding the solvent (water). Fritz and
Schlunder reported good agreement between experimental data and this model.
The IAS model was simplified by DiGiano et al. (67) into what has been
called the simplified competitive adsorption model (SCAM).
KINETICS OF ADSORPTION
The rate of adsorption is regulated by three steps: (i) transport of the
adsorbate from the bulk solution to the activated carbon particle (film diffu-
sion), (ii) diffusion of the adsorbate through the internal macro- and
transitional-pores of the carbon particle (internal diffusion), and (iii)
adsorption of the solute onto the carbon's internal surface (59). Within a
well-mixed batch reactor, internal diffusion is generally found to be rate-
limiting. The rate of internal diffusion may be determined by one .of two
parallel mechanisms: molecular diffusion through the fluid that fills the
internal pores (pore diffusion) or diffusion of adsorbed solute along the pore
walls (surface diffusion). The overall rate of internal diffusion is given by
the sum of the rates of transport by the two mechanisms. If the rates are
widely different, the overall transport is given approximately by the rate of
the faster of the two parallel processes.
Pore diffusion into a sphere, if it is assumed that diffusion occurs
radially inward, is characterized by the continuity equation (68):
.9C = DrA_c
at UL 2
, 2
H
(12)
3r
r 9r
20
-------
where C is the concentration at any radius r for a given time t. The diffu-
sion coefficient D is assumed to be constant. If adsorption occurs, then
equation 12 must be modified to the form:
9C
_ T.
~ D
3r
_2 3CS
r 3r
]
3t
(13)
If it is assumed that adsorption is instantaneous and is further governed by a
linear isotherm of the form S = RC, (S = adsorbed concentration, R = partition
factor), then the quantity [D/(l + R)] can be substituted for D in equation
12. This means that when adsorption occurs, for each.amount of solute present
in the solution enclosed by a pore, there is R times that amount adsorbed
along the surface of that same pore. In its adsorbed state, the solute would
not directly contribute to a. driving force for diffusion in the bulk phase of
the pores. So the amount adsorbed (R), according to this reasoning, would not
be involved in establishing a concentration gradient. Consequently, it would
require (R + 1) amount of solute to achieve the same concentration gradient
that it took one amount of solute to achieve without adsorption (18). Equa-
tion 12 then becomes
_
at
3r
2 9C>
~ 7~J
r 3r
(14)
This expression has been solved analytically (68) for the case where the
total amount of solute in the sphere after time t is compared to the total
amount adsorbed at equilibrium. This method, which will be used in Section 7
of this report, has also been used by Roberts (69). Roberts successfully
employed this and other related equations to compare two dimensionless parame-
ters: the fractional approach to equilibrium, f = (C - Ct)/(C - C^) versus
dimensionless time, T = Dt/a . Using these two parameters, he was able to
model the rate at which a synthetic zeolite adsorbent adsorbed normal paraf-
fins from binary liquid solutions.
The diffusion equation (equation 12) has been discussed by Walker et al.
(13) in relation to the diffusion of gases into zeolite and carbon molecular
sieves. Walker et al. suggest two models to account for the effect of ad-
sorption. In the first, the gas held by the solid would be considered to be
in an occluded state. As such it would not behave as a free gas following
ideal gas laws. Rather, it would be affected by force fields which are sig-
nificant throughout the cross section of the micropores. Such.an occluded gas
would not be so firmly fixed in one position that it could not diffuse through
the system., Rather, it would undergo activated (hindered) diffusion, and its
driving force, 3C/3x, would be in units such as gram-mole/cm pores/cm dis-
tance.
In a second model, conceived by Walker et al. (13), it is assumed that
the gas within the adsorbent is partitioned into two phases: (i) those mole-
cules occupying the open porosity of the solid which would be relatively free
to diffuse (pore diffusion); and (ii) those molecules adsorbed in a layer on
the internal walls of the solid, which would be relatively non-mobile. Walker
21
-------
suggested that the Langmuir isotherm could be used to determine partitioning
between adsorbed and non-adsorbed gas within the pore matrix.
The first of these models, Walker suggested, would be more useful in
describing the behavior of molecular-sieve materials. However, the second
could be useful with porous carbons, if consideration were to be given to both
pore and surface diffusion. Walker further stated that when pores are of
near-molecular dimensions, the two models are physically identical.
^Both pore and surface diffusion were considered by Fritz, Merk, and
Schlunder (61) in their model of competitive adsorption of two dissolved
organics onto activated carbon. They found remarkable agreement between
experimental and theoretical kinetics data when they considered
ni,k Yi,p 3r
pk Yi,s 9r
(14)
where n. , is the flux of component i, J. D is t*ie Pore diffusion coefficient
of component k which at radius r has a pore concentration of x., Yj. is the
surface diffusion coefficient of component i which has a surface concentration
S within a grain with particle density pR.
Surface diffusion was proposed to be the predominant intraparticle mass
transfer process by Crittenden (70,71) in his model for the design of fixed-
bed adsorbers. He therefore based his diffusion equation on the surface phase
concentration, essentially the last term in Eq. 14. He based the numerical
solution for his differential equation on the solutions given by Crank (68).
Kinetic studies have also been presented by others, notably in the early
work of Weber and Morris (72), who experimentally evaluated the rate of
adsorption of several single solutes. In this article, they developed no
theoretical models to describe the data, but rather calculated a "rate coeffi-
cient," k, in units of moles solute adsorbed per gram carbon per square root
of hours.
ADSORPTION OF ORGANICS FROM WASTEWATER EFFLUENT
Considerations Regarding Composition
Real waters and wastewaters are complex in composition: they contain a
multitude of organic compounds, which differ greatly with respect to molecular
weight, chemical composition, functional groups, polarity, adsorbability, and
diffusivity. The presence and concentration of a limited number of these
organic compounds can be determined through advanced analytical procedures.
Only a small fraction of the organic material in natural waters and waste-
waters can be identified specifically as single substances with currently
available techniques.
22
-------
The molecular weight distribution of the soluble organic compounds in
secondary effluents, as determined by DeWalle and Chian (73), included 36%
with a molecular weight less than 100, 21% with molecular weight between 100
and 500, 28% with molecular weight of 500 to 5000, arid 15% with molecular
weight greater than 5000. Activated carbon treatment removed 57% of those
compounds with molecular weight < 100, 82% of those compounds with molecular
weight of 100-500, and 90% of, those compounds with molecular weight > 500.
Many of the compounds of molecular weight less than 100 were polar. The high
molecular weight compounds consisted of carbohydrates, proteins, and humic
acids. Similar molecular weight distributions were found by Parkin (74) and
Keller (75) for effluent from wastewater treatment plants, notably the Palo
Alto Water Quality Control Plant.
Wastewater Effluent Adsorption Behavior
Most fundamental kinetic and isotherm studies have been conducted on
single solutes or on a few well-characterized model compounds. Real systems
found at water and wastewater treatment plants have been predominantly studied
in an empirical way (76,77), i.e., data from these facilities have shown how
much of various compounds have been removed from-»the solution phase. Although
such articles have been useful, they have not usually been developed suffi-
ciently to offer fundamental, quantitative understanding of the behavior of
competing solutes.
Because of the complexity of real waters and wastewaters, those who study
their adsorption choose to tise collective parameters such as total organic
carbon (TOG), dissolved organic carbon (DOC), or total organic chlorine
(TOCL); settle for data on only a few important compounds; or resor't to a
combination of these approaches.
Such an attempt was made by Frick (78). He applied the simplified com-
petitive adsorption model (65) to unknown mixtures by idealizing the mixed
solutes as a semi-defined system having a smaller number of pseudo components.
He achieved this by adding a specific amount of a tracer substance to the
unquantified mixture. By watching the adsorption behavior of the added com-
pound in the presence of the unknown organic compounds, he felt that it was
possible to obtain information on the single-solute data and on the concentra-
tion of key components within this mixture.
Secondary effluent, characterized by TOG, was used by Hsieh (79) for
kinetic experiments over time periods up to 270 hours for a number of ratios
of activated carbon to wastewater volume. After 48 hours, the control concen-
tration in these experiments experienced a gradual but significant reduction,
probably due to bacterial decomposition. Clearly, caution must be exercised
in adsorption rate exeriments to avoid the complication of biodegradation
phenomena being superimposed on adsorption phenomena.
23
-------
SECTION 6
MATERIALS AND METHOD'S
CELLULOSE
Two reagent grade celluloses were used in this study. Cellulose M is a
microcrystalline cellulose prepared for column chromatography by EM Laborato-
ries, Inc. of Elmsford, New York. The ash content of cellulose M is 65 ppm
(parts per million) (26). Cellulose £ is alpha cellulose fiber (a-cellulose)
distributed by the Sigma Chemical Company, St. Louis, Missouri. The ash
content of cellulose Z is 0.19% (method described below).
LIGNOCELLULOSIC MATERIALS
Twenty lignocellulosic materials were selected for experimentation,
representing the three major taxonomic groups of terrestrial plants (softwood
trees, hardwood trees, and grasses), as well as three types of paper, several
varieties of fruit pits and shells, steer manure, and two types of decayed
woods (pecky cedar and cubic brown rot from a lodgepole pine). Table 1 (see
Section 7) lists the materials selected along with their composition, deter-
mined as described below. The materials were chosen with a view to obtain a
variety of lignocellulosics with widely ranging fractions of the tissue compo-
nents (lignin, holocellulose, extractives and ash). Table 1 indicates this
criterion was satisfied. Thus we expect our experimental results could be
applied to virtually any natural lignocellulosic material and even, perhaps,
to man-made waste materials composed largely of natural of modified lignocel-
lulosics .
Pecky cedar, a waste product from cedar milling, was donated by Califor-
nia Cedar Products, Stockton, California. Cubic brown rot was collected in
the Sierra Nevada from fallen lodgepole pine. The peat sample was supplied by
Dynatech R/D Co., Cambridge, Massachusetts; the peat had originated from
Minnesota and thus was likely to be the decay product of arborescent plants.
Steer manure was a commercially available garden supplement distributed by
Sequoia Chemical Corporation, Chino, California. Prune pits were.donated by a
local fruit processing facility. Walnut shells were gathered locally from
English walnut trees. Coconut shells were separated by hand from store-bought
coconuts. Newsprint paper was taken from roll ends purchased from a local
newspaper and was free of print. White fir wood was donated by the Forest
Products Research Laboratory, University of California at Berkeley. The other
wood samples were derived from naturally dried branches (free of bark), except
for walnut which had been kiln-dried. Computer paper was taken from waste
bins at the Stanford Center for Information Processing. The kraft paper
24
-------
sample was prepared from standard cardboard boxes used for shipping. Some of
the agricultural samples were provided courtesy of Lawrence Berkeley Labora-
tory, Berkeley, California while the others were obtained form various agri-
cultural sources in California.
REFUSE-DERIVED FUEL (RDF)
Refuse-derived fuel refers to the fraction of municipal solid waste
remaining after the raw waste is classified to remove metals, glass and other
inorganics. A sample of RDF was provided by the Occidental Research Corpora-
tion, LaVerne, California. The RDF was produced from municipal solid waste
from San Diego, California, at the Environmental Protection Agency sponsored
demonstration classification and flash pyrolysis plant in El Cajon, Califor-
nia. The complete classification scheme used at the plant is described in
detail elsewhere (80); briefly, it consists of pre-shredding, air classifi-
cation, drying and final screening. The RDF sample used in this work was dry,
relatively odorless, and visibly heterogeneous. Particles varied in size and
type from powdery inorganic grit to one to two inch pieces of paper and minor
amounts of plastic. The presence of relatively large amounts of grit indi-
cates the RDF did not receive final screening. Also visible were small pieces
of twigs and stalks, cloth, string, yarn, aluminum foil and aluminum metal.
The RDF was prepared for experimentation by grinding, as described below.
The ash content of the ground RDF was approximately 16%. However, the RDF was
then sieved prior to experimentation, resulting in a final ash content of.-.
approximately 12% and a particle size range of 74-500 ym. Klumb and Brendel
(81) report air,classified solid waste analyses for over 650 samples which
indicate an average ash content of 19.2% (53.8% maximum, 7.6% minimum).
Hence, the prepared RDF sample used in this work may be considered representa-
tive of relatively efficient, but not atypical, commercial classification ,
technology.
All other materials
MATERIALS PREPARATION
Cellulose M and cellulose £ were used as supplied.
were ground and sieved prior to use. '.'-.
Preparation for Grinding
All lignocellulosics except prune pits and steer manure were air^-dried
and cut or broken into small (1 inch) pieces prior to grinding. Prune pits,
which were taken directly from the food processing facility, had large amounts
of the fruit flesh adhering to them. The flesh and kernel were separated from
the pit coat by blending in hot water in a large, stainless steel blender.
Both the flesh and kernel were thus washed away and the pit coat broken to
small pieces. This was necessary to yield a relatively high lignin material,
which was desirable for our experimental purposes, and to ensure sample homo-
.geneity. The steer manure was also blended briefly in cold water and then
placed in a No. 30 (U.S. Standard) sieve (0.595 mm openings) and rinsed
25
-------
repeatedly with cold water to remove dirt and grit and thus reduce the ash
content. Both prune pits and steer manure were air-dried prior to grinding.
Grinding
All lignocellulosics were ground in a Wiley mill to pass a 0.5-mm screen.
The mill operates by a shearing action between rotating and stationary steel
knives. All materials except coconut shells', walnut shells and prune pits
ground very rapidly. The shells and pits, however, were very resistant to
grinding and caused the mill to heat up. To avoid thermal alteration of the
lignocellulosic, the mill was operated cyclically, to provide copl-down per-
iods and insure the temperature never exceeded 50-80°C.
Sieving
The ground samples were air-dried by placing them in direct sunlight to
speed the sieving process. Sieves were shaken for 10 to 20 min depending on
the type of sample. The fraction of the lignocellulosic sample passing U.S.
Standard sieve No. 45 (0.354 mm openings) and retained on U.S. Standard sieve
No. 200 (0.074 mm openings) was retained for analysis and experimentation.
This fraction is nominally composed of particles in the size range 74-354
ym. This size fraction was selected as a compromise between the need for
sample homogeneity, the resistance of some materials to grinding, and the
requirements of the analytical procedures discussed below.
Preparation of Subsamples
The last step in materials preparation was the production of 10-20 gram
subsamples of the 45200 mesh fraction using a standard riffle sampler.
Riffle samplers are designed to yield subsamples which are representative of
the sample as a whole (i.e., having the same particle size distribution and
overall composition). This step thus insures that the analyses of one subsam-
ple yield data applicable to the subsamples used in the pyrolysis experiments,
and further that replicate pyrolysis experiments are performed with the same
starting material.
RAW MATERIALS ANALYSIS
All lignocellulosics were analyzed using the methods described below to
determine their composition in terms of the basic cell wall constituents:
lignin, holocellulose, extractives and inorganics. Additionally all materi-
als, including the reagent celluloses, were analyzed as described below for
their carbon content.
Ash
The ash content is a measure of the inorganics present in the lignocel-
lulosic material, but is not necessarily quantitatively equivalent, due to
volatilization of some of the organics during the ashing procedure. The
method used in this study is an adaption of the method described by the U.S.
Forest Products Laboratory for determination of ash in wood (82). The
26
-------
procedure yields the percent residue (dry weight basis) upon ignition at 600°C
to constant weight. The analysis is carried out using porcelain crucibles, 3-
to 5-gram samples, and carbonization of the sample over a bunsen burner, fol- ,
lowed by. ignition at 600°C in a muffle furnace overnight (10 hours or more) .
Analyses were cqnducted in triplicate. Standard .deviations for the analysis
varied from less than 1% of the mean for homogeneous materials to 57% of the
mean for very low ash or nonhomogeneous materials.
Silica-Free Ash
Silica-free ash is a measure of inorganics other than silica present in a
lignocellulos'ic material. The analysis was performed by Ultrachem Corpora-
tion, Walnut Creek, California. Weighed portions of the sample ash (residue
upon ignition at 600°C overnight) were treated with sulfuric and hydrofluoric
acids, muffled at 1200°C to volatilize the silica, dessicated aad weighed.
The acid treatment and volatilization were repeated until constant residue
weight was achieved. Residue is defined as silica-free ash, but other com-
pounds not volatilized at 600°C could be volatilized under the acidic and
higher temperature conditions of this analysis. Standard deviation on repli-
cate analyses was approximately 2%.
Extractives
Extractives in lignocellulosics consist of materials soluble in neutral
solvents but not part of the lignocellulose matrix itself. Such materials are
resins, tannins, waxes, gums, fats, and phenolics. The method used is an
adaption of the Forest Product Laboratory's determination of extractives in
wood (82). Three to five grams of air-dried sample is accurately weighed into
a fritted glass extraction thimble. The sample is also weighed into a pair of
glass-weighing bottles for separate drying and determination of sample mois-
ture content. The thimble is placed in a soxhlet extraction apparatus and
extracted with a 2:1 benzene-ethanol azeotrope at a rate of not" less than four
siphonings per hour for 8 hours. The thimble is then placed in a crucible
holder attached to a vacuum flask and rinsed 4-5 times with ethanol, allowing
a minimum 5-min soak time in ethanol between vacuum rinses. The sample is
then rinsed twice with anhydrous ether to aid drying. The thimble is placed
in a vacuum oven at 100°C for approximately 36 hours and reweighed. Weight
loss corrected for original moisture content is reported as percent extrac-
tives. Analyses are run in triplicate. The standard deviation for the analy-
ses varies from 20% of the mean for a material with a very low extractive
content to less than 1% of the mean for extractive-rich materials. Most
sample standard deviations lie between 1% and 10% of the mean.
Lignin
Lignin is a high molecular weight polymer of aromatic alcohols which
interpenetrates the cellulose fibrils and, hemicellulose polymers and in es-
sence cements the lignocellulose matrix together. The method used is that
known as the acid-insoluble lignin - modified hydrolysis method (82). Extrac-
tive-free sample is subjected to sulfuric acid hydrolysis in two steps to
hydrolyze the carbohydrates, leaving the acid-insoluble lignin as a solid
residue to be captured on a filter and measured gravimetrically. Only air-
27
-------
dried, extracted sample is used since oven drying has been observed to in-
crease the apparent lignin content, due to alteration of the carbohydrates,
part of which are thus transformed into acid-insoluble form. The first hy-
drolysis step is accomplished in 72% BUSO* for one hour at 30°C (achieved by
placing the vials containing the samples in a water bath). The sample/acid
mixture is stirred three times during this period. The sample/acid mixture is
then diluted to a 4% I^SO^ solution for secondary, hydrolysis, which is accom-
plished by autoclaving for one hour at 121°C. The autoclave is allowed to
exhaust slowly enough to avoid boiling and evaporation of the liquid. The
acid insoluble lignin is then filtered onto glass fiber filters placed in
filtering crucibles. Crucibles are then dried, weighed, ignited overnight,
and reweighed. The weight .loss upon ignition is defined as acid-insoluble
lignin. Analyses are performed in duplicate with standard deviations for
lignin as percent of extractive-free sample generally less than 1% of the
mean. This value is then corrected, using the extractives content as de-
scribed above, to lignin as percent (dry weight basis) of the raw sample.
Holocellulose
Holocellulose is defined as the total carbohydrate content of the ligno-
cellulosic, which is composed of cellulose and hemicellulose. For the pur-
poses of this study, we found it sufficient to determine percent holocellulose
("cell^ by difference U-e., acell = 100 - a1± - a - a ,). This may
result in a positive bias of the estimate of tne holocellulose content, espe-
cially if there is a substantial portion of the native lignin which is acid
soluble and thus not measured in the lignin analysis. For example, up to 15%
of hardwood lignin may be acid-soluble (32), which, for a typical hardwood of
20% lignin, 75% holocellulose, and 5% extractives (negligible ash), would
result in an experimental estimate of holocellulose (by difference) which was
only approximately 4% too high (78% versus 75%).
Percent Carbon
The percent carbon (dry weight basis) of raw materials (and chars as
described below) was determined using a WR-12 Carbon Determinator, Model 761-
100, made by Leco Corporation, St. Joseph, Michigan. Approximately 75 mg of
dry sample (40 mg for chars) was accurately weighed (± 0.5 mg for samples,
± 0.1 mg for chars) into a special crucible and then covered with copper cata-
lyst and iron chips. The crucible was then purged with pure Oo and ignited in
an induction furnace for 70 seconds. The combustion gases were dried, passed
over a catalyst to convert all CO to C02, and through a molecular sieve to
retain C02« The sieve was next heated to release the C0? which passed through
a thermal conductivity detector. The integrated result for a sample was com-
pared to a standard curve derived from standards supplied by Leco. Analyses
were generally performed in triplicate. Standard deviations were between 1%
and 2% of the mean for raw materials, and between 1% and 4% of the mean for
chars.
28
-------
PYROLYSIS
The equipment used in the pyrolysis experiments is schematically illus-
trated in Figure 1. Ten to twenty gram dry samples were accurately weighed
and placed in either a 275-ml or 750-ml quartz glass reactor (previously
tared). The reactor top was sealed to the reactor bottom using high-
temperature silicone grease and metal clamps. The entire reactor assembly was
suspended over the furnace, which could be raised around the reactor or low-
ered with a heavy duty laboratory jack^ The reactor assembly was attached to
the argon purge gas system and the tar and condensible by-product traps with
silicone rubber tubing.
The argon purge gas system consisted of research grade argon (99.998%)
supplied through a two-stage regulator. The argon was first passed through a
12-inch column of silica gel to ensure dryness and then over copper filings
contained in a tube heated to 500°G to remove oxygen. Argon flow was regu-
lated by a Nupro "J" series miniature valve and measured with a rotameter
(Matheson Gas Products, Tube 600). the argon was introduced to the reactor
through a ground glass fitting in the reactor'top.
The furnace (5inch I.D. x 6 inches deep) was constructed from cylindri-
cal half section elements supplied by Thermcraft, Incorporated, Winston-Salem,
North Carolina. The furnace temperature was regulated with a temperature pro-
gramming subsystem consisting of a commercial furnace controller (Model 72-1,
Love Controls Corporation, Wheeling, Illinois) and a voltage generator made at
Stanford University. The programming system allowed controlled, linear fur-
nace-heating rates from 0.03 to 25 °C/min.
Temperature was monitored in the reactor and furnace by 1/16-inch diame-'
ter inconel-sheathed, grounded-junction Type, K (chromel-alumel) thermocouples.
The tip of the sample thermocouple was positioned approximately in the center
of the sample. Temperatures were recorded with a Leeds and Northrup Speedomax
250 multipoint recorder, calibrated for Type K thermocouples. The recorder
could monitor up to 10 signals simultaneously with time between successive
readings of 1 second. Chart speed could be varied from l/,4 inch/hour to 15 ",
cm/hour, selection being made on the basis of the expected heating rate.
Heating rates reported in this study were determined as the slope of the
temperature-time plot produced by the recorder. For the slowest heating rates
(~ l°C/min) the sample temperature followed the programmed furnace temperature
very closely, and sample heating rate was.linear throughout the experiment.
However, both the medium (~ 15°C/min) and high (> 100°C/min) rate experiments
resulted in non-linear sample temperature versus time plots, with higher rates
at lower temperatures (between 200-350°C). Since it is within that tempera-
ture range that the pyrolytic reactions of dehydration and depolymerization
begin, it is probable that the heating rate estimated for that temperature
range would be more closely related to char yield than would the average
sample heating rate estimated over the entire range (20°C to Tf > 500°C) or
any other portion of the temperature-time plot. Thus, the heating rates
reported herein are taken from the plots between 200-350°C for the medium rate
experiments and below 350°C for the high rate experiments. Note also that the
maximum controlled heating rate achievable with the furnaces used in this
29
-------
Q.
p
U. o
o.
o
v>
^ - o
«
Q.
o
o
ce
^ Si
o E «-
S 5 8
»ife8
lf^5
jo o> 2
0)
I-
to
H
CO
O
S-l
o
o
0)
,13
O
en
0)
VI
30
-------
study was approximately 25°C/min.' Consequently, to achieve very high heating
rates (> 100°C/min), it was necessary to immerse the reactor into the pre-
heated furnace. Hence, the maximum sample heating rate increased with the
final pyrolysis temperature.
The reactor top contained an integral tar trap which collected easily
condensible materials. The remainder of the gaseous and condensible by-
products and argon purge gas were routed via silicone rubber tubing to a
watercooled condensor and condensate trap, through a water seal to prevent
inadvertent entrance of atmospheric oxygen, and finally to an aspirator dis-
charging to the 'sewer.
i
Before a pyrolysis experiment began, the sealed reactor containing the
sample was purged with argon (> 250 ml/rain) for a minimum of one half hour to
remove atmospheric oxygen. The argon flow was then lowered to approximately
50 ml/min for the duration of the pyrolysis experiment. The sample was held at
the desired final temperature (500-900°C) for 1 hour, after which the furnace
was lowered rapidly from around the reactor and the reactor allowed to cool.
Argon flow was increased during the cooling phase to approximately 150 ml/min.
When the sample temperature dropped below 100°C, the reactor bottom and char
were transferred to a dessicator. Char was weighed in the reactor bottom and
the yield of char calculated on a dry-weight basis. Preliminary experiments
with lignocellulosics indicated for triplicate pyrolyses a standard deviation
>for char yield of 0.5-1% of the mean. Later experiments to elucidate char
yield of lignocellulosics were run in duplicate with the same low standard
deviation.
ACTIVATION
The equipment used in the activation experiments is schematically illus-
trated in Figure 2. Accurately weighed 1- to 3-gram char samples were placed
on the frit of the quartz-glass, gas preheating assembly and the 750-ml
quartz-glass activation chamber was fitted over the sample by means of a
ground joint. The activation chamber had a frit at the top to allow gases to
escape while retaining the char under activation and a ground joint to allow
introduction of a thermocouple (identical to those described previously) for
activation temperature measurement.
The entire activation assembly was connected via silicone rubber tubing
to the purge and activation gas system. The argon system described previously
was used to supply the inert purge gas. Carbon dioxide (99.9%) was used as
the activating (oxidizing) gas with no further treatment. A three-way valve
allowed virtually instant switching from argon to carbon dioxide and vice-
versa. Gas flow was measured with rotameters (Matheson Gas Products, Tubes
600 and 603).
The activation chamber assembly could be lowered into and removed rapidly
from the activation furnace (2-5 seconds). The activation furnace (5-inch
I.D. x 12 inches deep) was constructed in the same manner as the pyrolysis
furnace. Temperature was controlled within ± 5°C with a Love Model 72-1
temperature controller. Furnace temperature was sensed with a Type K
31
-------
RECORDER *
EXHAUST
i A, ,
1-2 THERMOCOUPLES
3 ACTIVATION CHAMBER
4-5 QUARTZ GLASS FRIT
6 Ar/C02 INLET
7 FURNACE
Figure 2. Schematic of activation equipment.
32
-------
thermocouple (identical to those described above) and activation chamber and
furnace temperature were recorded on the Leeds and Northrup multipoint
recorder described previously.
Activation experiments were conducted by purging the preweighed char .
sample in the activation chamber for 20 min with argon while the entire assem-
bly was outside the preheated furnace. The assembly was then lowered quickly
into the preheated furnace; purging with argon continued until the desired
.temperature was reached within the activation chamber. At this point the gas
valve was switched to admit carbon dioxide to the activation chamber and
timing by stopwatch was, begun. The gas flow rate for both argon and carbon
dioxide were previously determined in trial runs to assure fluidization of the
char samples at the activation temperature. The flow rate, of course, depen-
ded on the particle size and density of the char. For prune pit char activa-
tion, a C(>2 flow rate of 580 ml/min was used, whereas for RDF char a flow rate
of greater than 3 1/min was necessary to achieve marginally acceptable homoge-
neity of fluidization, due to the clumping tendency of the RDF. At the end of
the desired activation time (determined by iteration, since percent burnoff is
oC greater interest in this work than is activation time), the gas flow was
switched to argon and the reactor quickly removed from the furnace and allowed
to cool. When the interior temperature reached 100°C the activation assembly
was opened and the char transferred to a weighing bottle or boat. Complete
recovery was not possible, because some char visibly remained either on the
top frit or side walls of the activation chamber. Nevertheless, percent
burnoff for prune pit char had a standard deviation of only approximately 4%
of the mean for three activation replications.
CHARACTERIZATION OF CHAR AND ACTIVATED CARBON
Percent Carbon
The percent carbon (dry-weight basis) of lignocellulosic chars was deter-
mined as described above under "Raw Materials Analysis."
Surface Area and Micropore Volume
The surface and pore characteristics of the chars and carbons were de-
rived from gas adsorption isotherms determined with a Orr Model 2100 surface
area and pore volume analyzer (Micromeritics Instrument Corporation, Norcross,
Georgia). The adsorbate gases and adsorption temperatures used in this study
were nitrogen at 77K, and carbon dioxide at 195K. For most chars, adsorption
equilibrium was reached quickly (Ap/At < 1.33 Pa/min in 15 min), while for
some others, adsorption was not completed after 18 hours. Adsorption equilib-
rium in activated carbon analysis was generally reached within 30 min. Only
isotherms of experiments in which equilibrium was reached quickly (within
30-60 min) were evaluated quantitatively. The slower adsorption experiments
were used to infer qualitative results only.
Adsorption isotherms were interpreted using the Brunauer-Emmett-Teller
(BET) equation (Eq. 1) discussed earlier. In most experiments, the BET plot
was linear within a fairly narrow range of relative pressures (0.01 < p/p <
s
33
-------
0.10) and the data within that range were analyzed by linear regression for
the calculation of surface area. CEoss sectional areas of the adsorbed gases
were taken as follows: 16.2 x 10~20 m2 for N2 at 77K and 21.8 x 10~20 mZ for
COo at 195K (4). Saturation vapor pressures for the adsorbates were taken as
104 kPa (780 torr) and 105 kPa (787 torr) for N2 at 77 K (14) and C02 at 195 K
(83), respectively.
Pore size distributions were calculated from N2 (77K) adsorption iso-
therms using the Kelvin equation (Eq. 4) and the method of Cranston and Inkley
(12). This approach was quite useful for the activated carbons with a range
of pore sizes. However, the chars, with porosity largely limited to micro-
pores, did not adsorb much additional gas for p/pg > 0.4, which corresponds to
a pore radius of approximately 18 A for N2 adsorption at 77K. Consequently,
for chars we do not report pore size distribution.
For use in the modified Kelvin equation, we assumed the following values
for the constants applicable for nitrogen adsorption at 77K: surface tension
of 8.9 dyne/cm, monolayer thickness of 0.43 nm and a molar volume of the
liquid adsorbate of 34.9 cm3/mole (14).
Macro- and Transitional-Pore Volume
Macro- and transitional-pore volume was determined by mercury porosimetry
by the American Instrument Corporation. Powdered char samples (50-100 mg)
were analyzed in an Aminco mercury porosimeter and subjected to pressures up
to 60,000 psi (4.14 x 10 Pa). Absolute pressure and volume of mercury were
recorded simultaneously. Pore radii were calculated from intrusion pressure
using the Washburn equation (Eq. 6). Mercury-carbon contact angle was assumed
to be 130°. Results were corrected by the American Instrument Company for
temperature and for compression of mercury by subtracting the mercury penetra-
tion previously measured in the identical sample flask without porous solid
sample. No correction was made for compression or collapse of the sample.
Estimates of pore volume based on mercury penetration data to 60,000 psi do
not take into account closed porosity or volume in open pores less than 3 nm
in diameter.
MEASURING ADSORPTION OF ORGANICS FROM WASTEWATER EFFLUENT
Determination of Organic _Carbon
The dissolved organic carbon (DOC) of effluent samples was determined by
placing 3 to 10 ml of sample into an ampule, along with make-up water, K2S2Og
and 10% phosphoric acid reagents. In the presence of the acid, inorganic
carbon in the sample was converted to C0~ at room temperature. This inorganic
carbon was removed by purging the ampule contents with hydrocarbon-free oxy-1'
gen. The ampule mouths were then melted shut by an Oceanography International
ampule sealer. The sealed ampules were next autoclaved at 121°C for 4 hours,
during which time the organic matter in the sample was decomposed to C0?.
To determine the C02 present, the sealed mouth of the ampule was crushed
within a closed space that was connected by Teflon tubing to a Dohrmann D-52
34
-------
analyzer. The C(>2 was stripped from the ampule and delivered to the analyzer
by a stream of inert carrier gas. Within the instrument, the CO, was catalyt-
ically converted to CH*, and the GEL subsequently carried to a Flame loniza-
tion Detector; the ions were then collected by an electrode surrounding the
flame. The electrical impulse was amplified and detected by an electrometer.
By comparing the digital readout of an unknown sample with the readout from
known standards, one could determine the organic carbon content of the unknown
samples (84).
DOC Uptake Rate Experiments
Both adsorption rate experiments and the adsorption equilibrium isotherm
experiments were conducted with unchlorinated secondary effluent which was
collected from the Palo Alto Regional Water Quality Control Plant. This
effluent was first vacuum-filtered through a 0.45-pm millipore filter. Such
filtration effectively excluded any bacteria from the secondary effluent, thus
minimizing the risk of biological degradation Curing sorption experiments. If
not used immediately, the filtered effluent was stored at 4°C. The sample was
always used within 24 hours so that biodegradation would be minimized.
The isotherm tests were performed in a series of 250-ml erlenmeyer
flasks, one flask for each time interval analyzed (e.g., 1 rain, 2 min, etc.).
One set of flasks was used for each kind of carbon. Into each of these flasks
was placed a known amount of activated carbon together with 10 ml. of organic-
free water (prepared with a Milli-Q water purification system). The activa-
ted carbon samples were then degassed in a vacuum chamber for 30 min to ensure
that the internal pores of the carbon were completely wetted with the organic-
free water. This step was taken to ensure that the rate of solute uptake
during adsorption rate experiments would not be affected by the simultaneous
penetration of the internal porosity by the solvent (water).
Next 100 ml of filtered effluent was added, and the flasks covered and
shaken on a shaker table at.room temperature (20 to 25°C) for the specified
time. A wastewater sample without .any activated carbon was included as a
control.
Immediately after the specified time of adsorption, the wastewater sample
was forced through a rinsed 0.45-ym millipore filter to separate the activated
carbon from the wastewater. This step was included to ensure that the adsorp-
tion process would cease at the intended time and also that the activated
carbon would not be collected into the DOC ampule and contribute to the DOC
measurement. The pressure was applied to the water sample above the filter by
a syringe system (Luer-Lok, Becton-Dickinson and Company). It was found that
the millipore filter pads consistently leached 1.78 ± 0.11 mg DOC into these
samples; therefore, the pads were rinsed four times before use to minimize
contamination. The DOC of the water samples was determined in triplicate by
the method described above.
DOC Adsorption Equilibrium Experiments
DOC adsorption isotherm experiments were conducted with materials and
procedures much like those used for-the adsorption rate experiments.
35
-------
Unchlorinated secondary effluent was filtered, then introduced into erlenmeyer
reaction flasks together with a. known quantity of pre-wetted activated carbon.
The contents were shaken for a period of 24 hours at 20-25°C. Samples were
collected and prepared in ampules as described above and analyzed on th°e
Dorhmann DC-52 instrument.
However, in these experiments, the parameter of concern was the ratio of
activated carbon to volume of wastewater, rather than time. Moreover, for the
equilibrium isotherm experiments, reaction flasks were shaken for 24 hours to
assure that equilibrium conditions had been established. If longer equilib-
rium times had been chosen, biological or photochemical decomposition might
have played a significant role. System controls without activated carbon were
run to permit evaluation of the extent of DOC concentration decrease not
attributable to the presence of the adsorbent.
Ideally in adsorption isotherm experiments, the equilibrium concentration
of the adsorbate is varied over several orders of magnitude (66). However, in
our case this was not feasible. The upper limit for our experiments was
established by the concentration of DOC in real secondary effluent (10 to 14
mg/1); the practical lower limit was a result of the fact that between 0.5 and
1.0 mg/1 of the initial DOC was unadsorbable. These constraints limited to
one order of magnitude the range of concentration data which could be ob-
tained.
36
-------
SECTION 7
RESULTS AND DISCUSSION
PYROLYSIS AND ACTIVATION OF LIGNOCELLULOSICS AND REFUSE-DERIVED FUEL
Introduction
A wide range of both virgin and waste materials has been studied for the
production of activated carbon (1). However, much of the research has been
conducted by activated carbon manufacturers and is therefore proprietary and
unavailable to the public. The published research on pyrolysis and activation
for activated carbon production is mainly confined to coal and coal products^
There are some published results for pyrolysis and activation of lignocellulb-
sics (e.g., 2,3,85,86,87), but they are generally confined to specific educts
or sets of pyrolysis conditions. In general it appears that the results of
activation depend to some degree on the nature of the educt, its processing
history (carbonization) and the characteristics of the char (1,39,40). The
purpose of this section is to examine a wide range of lignocellulosics to
develop an understanding of the dependence of char yield and physical charac-
teristics on the educt composition and pyrolysis conditions. Additionally we
examine the activation of one lignocellulosic material in some detail.
Finally, refuse-derived fuel (RDF), prepared by classification of municipal
solid waste, is pyrolyzed and activated for comparison with results obtained
with the model lignocellulosics.
Char Yield as a Function of Lignocellulosic Composition
Representative subsamples of the 22 materials selected for this program
were prepared for analysis and experimentation as described in Section 6. The
44 pyrolysis experiments (duplicates for each material) were performed in
random order to avoid bias. Table 1 lists the results of the educt analyses
and pyrolysis experiments. The pyrolysis temperature was raised at approxi-
mately 15°C/min to 500°C, then held for 1 hour at that final value. The
heating rate was selected to allow comparison with the results of published
thermogravimetric studies of cellulose and lignocellulosics, generally conduc-
ted at 15°C/min (33,35,38). The final temperature was selected as the lowest
temperature at which the majority of pyrolytic decomposition and weight loss
.would be complete and significant porosity would have developed, based on
published results on cellulose and lignocellulosics (26,33).
As discussed previously, there is reason to expect (10,11) that char
yield could be predicted for a given set of pyrolysis conditions with an
equation of the following form
37
-------
S
13
rH
m
VO
-*-vt-COCOCO-*COCMCMCOCMCOCOCOCOCOCOCMCMCMCMrH
«*m>*COCO-*COCMCMCOCM CO CO COCOCOCOCMCMCMCMtH
rH CM VO
oo VD rH
rH CM *
VD O ON
CM co m
c^3 If} \Q \f} ^^ |I LO CO O^ *"CJ* i~H
l^in>d-COCOCOCOCMCMCOCM
m vo
o in
CO O
o oo
rH
01
? c*
o o
o o
TJ in CM cr» in CD o1* co r^« o
c^ e r4 c°^ c'c'Br§'§L^c^c^r^r^':^'B'§e:^'B
O CM COmf-CM«*CM CO
00 CM Ov
oo r^« 4J4JOJ^ficL,COCOCOCOCd 4JrHrH
C 0)
H 4->
C OJ
60 TJ
H
,rH CO
II CJ
H
60fl
OJ
~ CO
ca o
||
4-J rH
O 0)
Cd CJ
to
OJ cu
CO
rH O
S^
cd tH
4J CU
OJ CJ
OJ 0
CU JS
N
CU O
cu
B^ 4J
cd
II to
60
Xf -H
0)O rH
8 ,0 S
M
« cd i
^3 O
CO 4-1
Cd rH X
cd oj
OJ 4-1 8
OJ
to
<4-l
cd
O OJ
I
ca
cd
rH
OJ
H 8
rH -H
H ^-^ I
CO
CU O
t>9 CO O
O rH
cd
crjrH
M-(rH rH
COOJ rH
8 O OJ
«rH 8°
J3 O
CO ,C
cd cu
s-s a
t>S 0
II cu
II IH
rHCU
r*t ( ft] ;
co cum
cd a-H
8 8 TJ
a
o
H
4-1
cd
H
^
CU
TJ
TJ
cd
TJ
C
cd
CO
CO
cd
a
H
60
H
*
OJ
60
cd
£.
CJ
§
ca
CU
3
CO
CO
cd
o
H
"&
co
8«
I
o
CD-
CD
o
rH
I
TJ-
rH
cu
OJ
60
cd
IH
0)
4-1
OJ
rH
ft
cd
CJ
CJ
cu
4->
o
to
M-l
3
4->
cd
TJ
iH
cd
0
H
00
H
to
o
CU
o
to
pq
P4
OJ
CJ
3
o
CO
OJ
TJ
CU
a
OJ
4-1
OJ
TJ
4J
§
CO S TJ
>
OJ rH
£8
CU ft
CU
38
-------
(c + c, (a _ ) 1 a
A o 1 sfa ' '
i ,, + c0a, . + c_a + c. a
cell 2 lig 3 ext 4 ash
(15)
where
o>
c., = coefficients determined by multiple regression,
x = exponent to account for the increased char yield from holocellulose
believed to be caused by inorganics (x < 1), and
"cell' alio->
aext' aash
a £ = fractions of holoceilulose, lignin,
extracives7~ash7~and silica-free ash in the starting material.
The model assumes that the overall char yield is simply a sum of- the char
yields of the components and that there is no interaction between the compo-
nents during pyrolysis, except that certain inorganics (represented here by
silica-free ash) increase the char yield of the holocellulose fraction in a
way that can be modeled as a power function. This is the approach supported
by the work of Rothermel (37) who developed a model for char yield as follows:
Y -. [0.0917
0.624«
llg
0.285a
ext
(16)
Pyrolysis was accomplished in Rothermel's experiments in thermogravimetric
equipment at 15°C/min to 400°C. However, their model was developed in a
statistically unconvincing way. Furthermore, certain coefficients seemed
suspect based on our work with pure cellulose and reported values for lignin
char yield (32). Specifically, we expected the value of c in equation 15 t<
be approximately 0.20 rather than 0.0917 and the value of c.-^ to be approxi-
mately 0.55 instead of 0.624.
Subsets of our lignocellulosic composition and char yield data were ana-
lyzed by computerized multiple regression techniques. The results of these
analyses are listed in Table 2. Regressions were performed using either ash
or silica-free ash in the power function to test the hypothesis that silica-
free ash is more fundamentally related to catalysis of holocellulose char
yield. For materials for which silica-free ash had not been determined (i.e.,
those with ash contents lower than 2%), we assumed silica-free ash to be equal
to ash content for the purposes of the regression analysis. The decision to
limit the silica-free ash data was based on the high cost of commercial analy-
sis. Initial regression analysis with incomplete silica-free ash values were
expected to indicate whether, in fact, using silica-free ash in the power
function resulted in a better fit of the model to the experimental data. Had
this been true, additional silica-free ash analyses would have been warranted.
As discussed below, however;, this was not the case.
The multiple regressions summarized in Table 2 all set the power function
exponent equal to 0.462, a value taken from the work of Rothermel and others
(36,37) on the effect of fireretardant additives on cellulose char yield.
This assumption was made to allow initial analysis of the data by standard
linear multiple regression techniques.
39
-------
TABLE 2. SUMMARY OF RESULTS OF MULTIPLE REGRESSIONS VARYING THE
DATA UTILIZED AND THE INORGANIC PARAMETER IN THE POWER FUNCTION
Materials
Included
Code
la
Ib
Ic
in Analysis
All
22
Parameter
Used in
Power
Function
aash
asfa
None
Regression Coefficients
0
0
0
C0
.185
.192
.192
0.
-0.
set
cl
0722
0001
= 0
C2
0.540
0.529
0.529
C3
0.411
0.456
0.456
a
C4
0.97
1.06
1.06
Multiple
Correlation
Coefficient
(r2)b
0.9982
0.9981,
0.9981
Materials with
Ila
lib
aash <
asfa <
: 11.61
: 7.90
&
aash
asfa
0.188
0.192
-0
-0
.0025
.0613
0.551
0.551
0.368
0.374
1.16
1.24
0.9978
0.9978
III
Materials with
asfa < 5'25
*sfa
0.188 0.0059 0.542 0.411 1.17 0.9979
Materials with
IV asfa < 4.13
ex
sfa
0.188 0.0059 0.541 0.419 1.19 0.9978
Experimental data is fit by multiple regression to a model of the following
form
Char Yield = {CQ + c1(inorganic parameter)°'462}acell + C2alie
+ C3 aext + C4 aash
where a is expressed as a fraction (i.e., a < 1).
Analysis of variance allows computation of the multiple correlation coeffi-
cient (r ), adjusted for the sample size.
All materials were included in the regressions coded la and Ib in
Table 2. It is apparent from the values of r listed in the table that, for
the full set of materials, using either aagh or asfa in the power function
results in a very good fit of the model to the data (r > 0.99). Furthermore,
use of agfa in the power function results in a value of c, which is statisti-
cally indistinguishable from zero (confidence limits contain zero). This re-
sult is particularly interesting in light of the expectation that silica-free
ash would be a better indicator of inorganic catalysis of carbohydrate pyroly-
sis than total ash content. Had this expectation been correct we would have
anticipated c^ > 0.0722, the coefficient derived when total ash content was
used in the power function. Thus it appears that the model could be simpli-
fied to a simple linear additive form without any loss in predictive ability
(i.e., eliminate the power function altogether). The analysis coded Ic in
Table 2 is a separate regression for this simple linear model (setting c-, = 0)
which verifies this conclusion.
40
-------
There are two possible reasons for the apparent negligible effect of
inorganics on carbohydrate char yield: 1) the catalytic effect is thought to
saturate at relatively low ash contents (5-7% or lower) (35,36), whereas some
of our materials have ash contents as high as 17.9%; and 2) the assumed power
function exponent (0.462) may be too high.
Regressions IIA, lib, III, and IV (Table.2) were performed to examine the
possibility that the effect of inorganic catalysis was underestimated due to
the inclusion of materials with high ash contents. By restricting our analy-
sis to materials of successively lower ash contents as in II, III, and IV, we
might expect the catalytic effect to become more observable as evidenced by an
increase in coefficient CT determined by the regression. This, however, was
not observed. In fact, due to the large standard deviations of coefficient c,^
as determined in analyses II through IV, statistically the coefficients cannot
be distinguished from zero (confidence limits for c-^ contain zero).
To examine the possibility that the assumed power function exponent was
too high, we analyzed the full set of data with a nonlinear multiple regres-
sion technique to generate a new estimate for the exponent. This resulted in
a slightly higher estimate (~ 0.48) than originally assumed, while confidence
limits for c-, again contained zero.
These results favor the acceptance "of the char yield model Ic in Table 2,
which assumes no detectable effect of inorganics on holocellulose char yield.
Examination of the other coefficients lends additional support to the model.
Coefficient c indicates a holocellulose char yield for pyrolysis at 15°C/min
to 500°C of 19.2% which is consistent with Brunner's results for pure cellu--
lose in similar experimental equipment (26). Coefficient c~ indicates a
lignin char yield of 52.9%, which is within the range of Sarkanen's observa-
tions (32) and thermogravimetric results of Shafizadeh and McGinnis (33).
Coefficient c. = 1.06 indicates that slightly more inorganics are recovered in
the pyrolytic char than are measured by the standard ash analysis (ignition
overnight at 600°C). Considering that the ash analysis is,carried out at a
higher temperature, for a longer time, and in the presence of atmospheric
oxygen, it is quite reasonable that ash by ignition would underestimate inor-
ganic recovery in a char pyrolyzed in an inert atmosphere to 500°C.
Coefficient GO = 0.456 indicates the char yield from extractives to be
higher than that predicted by Rothermel's model (0.285). Rothermel's value
was derived from work by others (38) on thermogravimetric analysis of extrac-
tives from six lignocellulosic "fuels" (generally wood, stems, and leaves). A
possible problem is that the compounds measured by the extraction process vary
widely with the plant material and include oils, waxes, resins, tannins, gums,
phenolics, and terpenes. It is likely these compounds vary somewhat in their
pyrolytic char yield and therefore we might expect coefficient c^ to be sensi-
tive t;o the materials used in its derivation. Since our data base includes a
much larger sample of lignocellulosics, it is reasonable to assume that our
value !is more generally applicable. . .
In conclusion, the results presented here suggest that lignocellulosic
char yield for pyrolysis at 15°C/min to 500°C can be predicted satisfactorily
with the model listed in Table 2 as Code Ic. The fit of the model's
41
-------
predictions to the experimental char yield data is displayed graphically in
Figure 3. The spread of the measured values around the predicted values is
small, and the variance appears to be independent of the predicted yield.
There is no evidence of deviation for low-lignin, high-cellulose materials
(low predicted yields), nor for high-lignin, low-cellulose materials (high
predicted yields)
As a test of the ability of the above model to predict char yield, we
performed duplicate pyrolysis experiments on refuse-derived fuel (RDF). RDF
is prepared from municipal solid waste and therefore would be expected to be
quite heterogeneous, as confirmed in Section 6. These complications notwith-
standing, the RDF was subjected to exactly the same composition analysis and
pyrolysis conditions as were the materials used to derive the char yield
model.
Table 3 lists the results of the composition analysis and the pyrolysis
experiments. Also listed is the char yield predicted by the model on the
basis of the composition. The model thus appears to overestimate the yield.
However, the 95% confidence interval for the model's prediction is [40.5,
34.4] compared to the experimental value of 34.2%, whose 95% confidence limits
are [35.3, 33.1]. Thus there is reasonable agreement of experimental and
predicted yields, considering the assumption that RDF could be treated like a
natural lignocellulosic material.
In fact, RDF contains plastic, which is not taken into account in the
above prediction. Because plastics are not soluble in neutral solvents or
subject to acidic hydrolysis, the plastic content of RDF will be measured as
lignin in the composition analysis. By hand-sorting of a 100-gram subsample
of unground RDF, its plastic content was estimated to be approximately 4.9%
(dry weight basis), which is within the range expected for glass and metal-
free refuse (88). The pyrolytic yield of plastics or synthetic polymers is
known to vary with the polymer structure (89); e.g., polystyrene and polyethy-
lene are almost entirely volatilized, PVC (polyvinyl chloride) char yield is
on the order of 25%, and some polymers used as binders in pelletized activated
carbon manufacture have yields as high as 50-60%.
If we assume that the true lignin content of RDF lignin is given by the
measured value (15.75%) less the 4.9% plastic content, then the best estimate
of RDF's lignin content- is 15.75 - 4.9 = 10.85% lignin. Further assuming that
TABLE 3. COMPOSITION AND CHAR YIELD OF REFUSE-DERIVED FUEL (RDF)
Composition (%)
Char Yield
Pyrolysis at 15°C/min to 500°C
*cell alig
a
'ext
a
'ash
Predicted by
Model
Experimental
Mean
Std. Dev.
61.91 15.75 10.66 11.68
37.46
34.18
0.12
42
-------
60
50
LU
-30
O
UJ
o: 20
ZD
CO
<
UJ
0
CHAR YIELD MODEL
cELL+0-529aL.G
Y =0.192 < ,+ 0.529« ,+0.456 aEXT+1.06 ^ ///
95% Confidence Interval
o
IO 20 30 40
PREDICTED YIELD (%)
50
60
Figure 3. Measured yield versus yield predicted by char yield model for
pyrolysis at 15°C/min to 500°C
43
-------
the plastic is equally divided between polyethylene and PVC-type plastics, our
char yield prediction is revised to approximately 35.46% for RDF, which agrees
well with the measured value of 34.2%. This further supports the predictive
ability and wide applicability of the char yield model as presented.
Char Yield as a Function of Pyrolysis Conditions
To evaluate the dependence of char yield on pyrolysis conditions, a
series of experiments was performed with a small subset of lignocellulosics,
chosen to represent the full range of compositions encountered. Pecky cedar
was selected as a very high-lignin, low-ash material. Redwood served as a
moderately high-lignin, low-ash sample, representative of softwoods. Corn
stover was included as representative of agricultural wastes, having a rela-
tively low-lignin, high-ash composition. Computer paper was selected as rep-
resentative of finished paper products and also because it is composed almost
entirely of cellulose and inorganics. Thus the subset of lignocellulosics
used in this portion of the study included the extremes of composition likely
to be encountered in waste materials, as well as samples of more common compo-
sition. This approach was taken to highlight any effect that educt composi-
tion might exert on the relationship between pyrolysis conditions and char
yield (also char physical characteristics as discussed later).
The pyrolysis conditions varied in this experimental program were heating
rate () and final pyrolysis temperature (Tp). The heating rates studied were
l°C/min, 15°C/min and the maximum heating rate achievable with our equipment
( > 100°C/min and dependent on TF as discussed in Section 6). Pyrolysis was
conducted at each of the heating rates to final pyrolysis temperatures of
500°, 700°, and 900°C. Additional pyrolysis experiments to 600°C were conduc-
ted at 15°C/min.
Results of the pyrolysis experiments under varying conditions are listed
in Table 4. Standard deviations of the yield for duplicate experiments were
again on the order of 1 to 2% of the mean. To allow comparisons between the
materials using the original volatile solids content as a measure of potenti-
ally pyrolyzable organic material, ash-free yields were calculated as indi-
cated in Table 4. Ashfree yield is simply a comparison of the oxidizable
mass present before and after pyrolysis and is calculated using ash data
derived in a separate set of analyses reported in Table 5. Ash contents were
determined by ignition in a muffle furnace to the indicated temperature. For
materials with &asi, less than 1% (600°C), we assumed a . at higher tempera-
tures to be unchanged. For such low ash contents the correction from1yield on
total weight basis to ash-free yield is small enough such that refinement of
aash va-'-ues at T > 600°C would have insignificant effect on the calculation.
As an alternative method of comparing pyrolytic behavior of the materi-
als, the yield of carbon was calculated using the results of percent carbon
analyses of the educts and chars. The results of the analyses and the calcu-
lated carbon yields are presented in Table 6. Percent carbon data were not
obtained for the high heating rate chars. Neither percent carbon data nor
carbon yields are adjusted to account for the carbon which may be present in
the inorganic fraction of the educts, e.g., as carbonates, and possibly lost
during pyrolysis (e.g.,
' 44
-------
TABLE 4. SUMMARY OF PYROLYSIS OF SELECTED LIGNOCELLULOSICS FOR
VARYING PYROLYSIS CONDITIONS
Final Heat Rate3
Temp. (°C/min), $
Oo
Material
u
TF
Exp . 1 Exp . 2
Exp.l
Yield
%
Exp. 2
Ave.
s
Ash-Free
Yield, %b
Ave.
sc
A. Low Heating Rate
Redwood
Corn
Stover
-4
Computer
Paper
Pecky
Cedar
Sigma
Cellulose
500
700
900
500
700
900
500
700
900
500
700
770
900
700
B. Intermediate
Redwood
Corn
Stover
Computer
Paper
Pecky
Sigma
Cellulose
500
600
700
900
500
600
700
900
500
600
700
900
500
700
900
700
' 0.99
0.95
0.97
0.99
0.96
0.98
0.96
0.98
0.97
1.06
0.99
1.02
, 0.98
0.99
Heating
16.5
15.2
18.5
18.3
15.9
15.9
19.0
18.5
16.0
15.9
17.7
.. 15.5
17.5
18.3
17.0
14.3
0.91
0.98
0.99
0.98
0.99
1.00
0.96
1.01
.
'
Rate
16.8
19.2
17.5
17.7
18.2
17.9
15.7
17.3.
15.0
17.8
18.4
17.2
37.93
32.58
30.38
33.84
30.82
S.
28.63
33.58
29.86
27.05
50.38
43.51
41.96
40^.10
24.63
33.23
29.39
28.01
27.16
31.56
29.71
28.62
27.94
27.14
24.92
24.25
23.50
48.94
42.25
40.42
17.94
37.46
32.45
30.28
33.64
30.59
28.21
33.45
'
26.83
'
34.10
27.97
27 ,02
31.36
28.74
28.13
26.91
24.43
23.45
49.50
42.27
40.59
37.70
32.52
30.33
33.74
30.71
28.42
33.52
29.86
26.94
50.38
43.51
41.96
40.10
24.63
33.67
29.39
27.99
27.09
31.46
29.71
28.68
28.04
27.03
24.92
24.34
23.48
49.22
42.26
40.51
17.94
0.33
0.09
0.07
0.14
0.16
0.30
0.09
-<-
0.16
-
0.62
0.03
0.10
0.14
0.08
0.13
0.16
0.13
0.04
0.40
0.01
0.12
37.60
32.41
30.23
29.58
26.50
24.47
28.77
24.88
21.84
49.94
43.01
41.44
39.57
24.48
33.57
29.28
27.88
26.98
27 .17
25.30
24.35
24.06
21.82
19.56
18.97
18.13
48.77
41.75
39.98
17.78
2.53
2.16
2.02
0.24
0.40
0.84
0.15
0.21
,
2.32
1.86
1.80
0,22
0.35
0.79
0.16
0.11
0.14
0.68
0.47
0.47
:
TABLE 4 cont.
45
-------
TABLE 4 cont.
Final Heat Ratea
Temp. (°C/min), $ -
O/-1
Material
vj
TF
Exp . 1 Exp . 2
Exp.l
Yield
Exp.2
Ave.
s
Ash-Free
Yield, %b
Ave.
sb
C. High Heating Rate
Redwood
Corn
Stover
Computer
Paper
Sigma
Cellulose
500
700
900
500
700
900
500
700
900
700
aSee discussion
bAsh-free
yield
105
200
333
96
228
405
115
280
' 425
183
in text.
(% Yield -
nnn - ,
105
270
110
235
105
d
375
T
aash
re . ^
28
22
20
29
26
25
22
19
17
12
)100
.18
.24
.68
.56
.58
.35
.68
.52
.50
.17
27.95
22.25
29.13
26.33
22.80
18.99
17.20
where a^£
28.07
22.25
20.68
29.35
26.46
25.35
22.74
19.26
17.35
12.17
- 7
ih ~ /0
0.16
0.01
0.30
0.18
0.08
0.37
0.21
__
ash of
27
22
20
24
22
21
17
13
11
12
.96
.13
.56
.91
.00
.23
.23
.53
.58
.00
1.87
1.48
0.30
0.34
0.10
0.26
0.17
_._
raw material
as measured by ignition to T°C. See discussion in text.
GStandard deviation of ash free yield estimated by the following expression:
S
where Y
.J
AFY AF
average ash-free yield (%), Y = average yield (%),
standard
deviation of yield (%)-, «as^ = average percent ash of educt (%), and
SA - standard deviation of ash (%).
Heating rate unknown since sample thermocouple was accidentally disconnected
from recorder for first portion of experiment (< 400°C).
MgCOo ,-nn > MgO + CO-
J j3\j \jf £
(90)). Calculations which assumed the entire change in ash weight (Table 5)
was due to carbonate volatilization indicated that the necessary correction to
carbon yield was less than 1% of the uncorrected carbon yield and therefore
justifiably ignored.
Figure 4 displays ash-free yield as a function of final pyrolysis temper-
ature (Tp) for the low and medium heating rates. Linear regressions of the
data are used, since the plotted data had been shown to be statistically in-
distinguishable from the linear regressions (Appendix A). Similarly, Figure
5 displays ashfree yield as a function of Tp for the high heating rates,
46
-------
TABLE 5. DEPENDENCE OF ASH CONTENT OBSERVED UPON IGNITION
TO VARYING FINAL TEMPERATURES
Ash Content (c£gh)
Material
Computer Paper
Corn Stover
Redwood
Pecky Cedar
Sigma Cellulose
T =
Ave. %
6.66
5.90
0.15
0.88
0.19
600 °C
na
3
3
3
3
3
T = 700°C
s Ave. % n s
0.03 6.63 4 0.01
0.04 5.72 . 4 0.08
0.01
0.01
0.02
T = 900°C
Ave. % n s
6.53 4 0.05
5.23 4 0.17
JJ
n = number of analyses.
s = standard deviation in same units as average.
which are influenced by I for reasons described in Section 6. Figure 6 shows
carbon yield as a function of Tp, again using linear regressions of the data
for the same reason (Appendix A). It should be noted that the "regression" of
pecky cedar data (<|> = l°C/min) contains only two points, but is included none-
theless for purposes of comparison.
Figures 4 and 6 indicate that for a given heating rate, the slope of the
yield versus Tp plots (Tp > 500°C) are essentially the same for the range of
lignocellulosics studied. Thus it appears that char yield for pyrolysis of
any lignocellulosic is largely determined at temperatures below 500°C, the
weight loss above 500°C being relatively independent of educt composition, at
least for a given heating,rate. This is consistent, of course, with the fact
that the major pyrolytic weight loss occurs in the range 225-500°C for each of
the major lignocellulosic components (34).
Somewhat surprising is the observation that the slope of the plots is
steeper for the lowest heating rate, both for ash-free and carbon yields.
Thus the char yield improvement resulting from slow pyrolysis below 500°C is
somewhat eroded by continuing the slow heating rate to higher temperatures.
This is particularly evident in*the case of corn stover which, despite the
increased char yield (ash-free or carbon) for slow pyrolysis to 500°C, has
virtually identical yield for pyrolysis at medium and low heating rates to
900°C. A similar effect is seen for pecky cedar (ash-free yield) and computer
paper (carbon yield). This effect may be due simply to the greatly increased
time of exposure to the higher temperatures for the low-rate pyrolysis. The
results suggest, furthermore, that maximization of char yield for lignocellu-
losic pyrolysis may involve heating at different rates through different
temperature ranges: low rate below 500°C followed by more rapid heating to
the final desired temperature.
47
-------
TABLE 6. CARBON CONTENT AND CARBON YIELD FOR VARYING PYROLYSIS CONDITIONS
Percent Carbon
J. /Tl
Material
Redwood
Redwood
Corn
Stover
Corn
Stover
Computer
Paper
Computer
Paper
Pecky
Cedar
Pecky
Cedar
RDF
aPercent
^Standard
C0nly one
v/ *F Educt
°P Am-fri
L*/ win
°C Ave.
1/500 53.69
1/700
1/900
15/500 53.69
15/600
15/700
15/900
1/500 45.03
1/700
1/900
15/500 45.03
15/600
15/700
15/900
1/500 40.39
1/700
1/900
15/500 40.39
15/600
15/700
15/900
1/500 58.32
1/700
1/900
15/500 58.32
15/700
15/900
15/500 45.27
15/900
Carbon Yield - YC =
so
0.38
11
ti
0.38
"
"
0.99
11
'"
0.99
M
0.41
"
ft
0.41
0.35
tt
"
0.35
11
tt
0.04
**
Cc/Co
Char (Cc)
Ave.
87.94
93.50
99.05
86.96
93.05
97.47
98.68
74.79
76.65
75.07
70.62
72.73
75.45
76.15
69.39
75.04
72.82
65.53
70.87°
70.94°
68.00°
86.55
94.49
84.00
92.29
96.75
61.72
65.07
(Y).
sc
0.65
0.71
0.81
1.93
0.89
1.44
1.06
0.64
1.15
1.23
1.17
1.24
2.81
1.19
1.26
1.11
0.93
0.73
2.06
1.06
1.63
0.04
1.95
0.47
2.11
Percent Yield
Gross Wt. (Y)
Ave.
37.70
32.52
30.33
33.67
29.39°
27.99
27.09
33.74
30.71
28.42
31.46
29.71°
28.68
28.04
33.52
29.86°
26.94
27.02
24.92°
24.34
23.48
50.38°
43.51°
40 . 10°
49.22
42.26
40.51
34.18
30.60
SY
0.33
0.09
0.07
0.62
.
0.03
0.10
0.14
0.16
0.29
0.14
0.09
0.13
0.10
0.16
0.16
0.13
0.04
0.40
O.Q2
0.12
0.12
0.24
Carbon (Yc)a
Ave.
61.75
56.63
55.95
54.53
50.94
50.81
49.79
56.04
52.27
47.38
49.34
47.99
48.05
47.42
57.59
55.48
48.57
43.84
43.73
42.75
39.53
74.77
70.49
70.89
66.88
67.20
46.60
43.98
H
0.83
0.61
0.62
1.62
0.61d
0.83
0.67
1.34
1.42
1.39
1.38
1.34d
2.08
1.30
1.21
1.00d
0.84
0.71
0.44d
0.49d
0.41d
1.84d
0.90d
1.55
0.40
1.43
0.39
1.47
deviation of YC estimated by
s = YC[(S0>
experimental value.
Estimated by setting missing
'CQ)2 + (SC/CC)2 + (SY/Y)2]1/2.
s values (s)
= 0.
48
-------
cr
Q
_l
UJ
o:
o
ui
UJ
a:
u.
i
x
CO
50
40
30
20
10
I PECKY CEDAR
k 2 REDWOOD
3 CORN STOVER
4 COMPUTER PAPER
/min
500
700
900
FINAL TEMPERATURE (°C )
Figure 4. Ash-free char yield versus final temperature:
heating rates
low and medium
49
-------
Q
UJ30
x20
O
"JJ 10
LiJ u
cr
0
i
-A
I REDWOOD
2 CORNSTOVER
3 COMPUTER PAPER
'N/350°C/min
50O 700 900
FINAL TEMPERATURE (°C )
Figure 5. Ash-free char yield versus final temperature: high heating rates.
50
-------
r-V
70 - I -
60
50
Q
U 40
30
O
CD
< 20
O
10
2
4
3
2
3
4
I PECKY CEDAR
2 REDWOOD
3 CORN STOVER
4 COMPUTER PAPER
-<£,vl50C/min
500
700
900
FINAL TEMPERATURE (°C)
Figure 6. Carbon yield versus final temperature.
51
-------
The plots in Figure 5 illustrate a similar point. The slopes are similar
to the medium heating rate while the absolute yields are much lower. Thus the
extremely high heating rates seem to exert their prime effect in depressing
char yield at temperatures below 500°C. This effect is even more pronounced
when it is considered that the experimental heating rates also increase with
the final pyrolysis temperature, as described previously.
Figure 7 presents ash-free yield as a function of the logarithm of the
heating rate. The data are represented as log-linear regressions for the same
reasons cited previously and also to allow comparison with Brunner's estab-
lishment of a log-linear dependence of cellulose char yield on heating rate
(26). The data were analyzed both by linear and log-linear regression, the
results of which are included in Table 7. Figure 7 presents the log-linear
regressions for the lignocellulosics examined in this section of the study, as
well as Brunner's results for Cellulose M, given by the following (26):
Yield =
In
+ a
where
2.4,
TABLE 7. LINEAR REGRESSIONS OF ASH-FREE CHAR YIELD VERSUS HEATING RATE
OR ln(HEATING RATE) FOR PYROLYSIS TO FINAL TEMPERATURE Tw
Material
Linear
Regression Constants'
Log-Linear
Regression Constants
Redwood
Corn Stover
Computer Paper
Sigma Cellulose
500
700
900
500
700
900
500
700
900
700
-0.08
-0.04
-0.03
-0.03
-0.02
-0.01
-0.09
-0.03
-0.02
-0.05
36.4
30.5
28.9
27.9
25.6
24.3
26.2
21.3
20.2
21.7
0.913
0.828
0.872
0.719
0.827
0.975
0.755
0.676
0.884
0.770
-1.99
-1.86
-1.58
-0.99
-0.82
-0.49
-2.44
-1.96
-1.72
-2.39
38.0
32.7
30.7
29.7
26.6
24.8
28.6
24.7
22.2
24.4
0.947
0.985
0.954
0.985
0.992
0.784
0.999
0.999
0.987
0.999
aModel format: Ash-free char yield (%) = a-i + an, where ty is expressed in
°C/min. . .
format: Ash-free char yield (%) = a-i(ln cj>) + aQ, where <|> is expressed
in °C/min.
52
-------
CO
$-1
a*
H
4J
CO
OJ
CD
3
CO
T3
rH
0)
V,
0)
0)
M-l
r^
(U
=1
W)
rl
33dd-HSV
53
-------
aQ - empirical constant = f(educt, Tp)
= 27.9 for Cellulose M pyrolyzed to 540°C
=25.3 for Cellulose M pyrolyzed to 710°C
and no correction to ash-free yield is necessary since the ash content is low
(65 ppm) .
Comparing Brunner's cellulose results with the lignocellulosic results in
Figure 7 and Table 7, we can make the following observations. As expected,
lower heating rates result in higher char yields for all materials investi-
gated. Most materials behave qualitatively similarly to cellulose, in that
there seems to be a log linear dependence of ash free char yield on heating
rate. However, slopes of the log-linear regressions vary widely (0.5 to
2.4) and in general, indicate a less marked dependence of lignocellulosic
char yield on heating rate compared to the cellulose results. In some cases
(redwood to 500°C, corn stover to 900°C), the data are not well suited to a
log-linear model. In fact, corn stover pyrolyzed to 900°C is better described
by a simple linear model.
The behavior of computer paper warrants discussion. Computer paper is
composed almost entirely of holocellulose and ash (a^g and ot t both less
than 1%). The plot of ash-free char yield (500°C) versus for computer paper
agrees closely with that of pure cellulose (540°C) in Figure 7. The higher
temperature of the cellulose data could explain the somewhat lower yields
observed and thus, the similarity of the computer paper and cellulose results
is even more striking. This is somewhat surprising in light of the hypothesis
that ash content catalyzes dehydration below ~ 240°C prior to the onset of
rapid depolymerization, which results in higher char yields. If ash constitu-
ents have the same effect, then cellulose in the presence of ash (e.g., com-
puter paper) would be expected to have a reduced dependence of yield upon
(i.e., a lower slope). For computer paper pyrolyzed to 500°C, this does not
appear to be the case, although such behavior is observed for computer paper
pyrolyzed to higher temperatures. The behavior of computer paper pyrolyzed to
500°C suggests either that its inorganic fraction does not contain any com-
pounds capable of catalyzing char yield of the cellulose or that the cellulose
present in the paper is not susceptible to chemical catalysis, perhaps due to
modification during the papermaking process (e.g., delignification and bleach-
ing) . Why the effect of heating rate decreases with increasing final pyroly-
sis temperature, however, remains unclear.
Corn stover exhibits the most unusual behavior in Figure 7, having a
significantly lower dependence of char yield on heating rate for all final
temperatures. This same behavior was also evident in Figure 5 and reflected
in Figures 4 and 6. This behavior may be an indication of the catalytic
effect of inorganics on pyrolytic char yield. Corn stover is a high carbohy-
drate material (71% holocellulose) with relatively high contents of ash (5.9%)
and silica-free ash (4.1%), and thus might be expected to exhibit such an
effect. As was the case with computer paper, the effect of heating rate
declines with increasing !.
54
-------
Lignocellulosic Char Properties
Surface areas of lignocellulosic chars produced under varying pyrolysis
conditions were determined by carbon dioxide adsorption at 195K using the BET
equation (Eq. 1). The results of the analyses are presented in Table 8, along
with calculated values for surface area expressed per gram of carbon and per
gram of educt. Surface area of high heating rate chars were not determined
since low values had previously been observed for cellulose pyrolyzed at such
rates. Most surface area data listed in Table 7 are based on a single deter-
mination. Replicate analyses were performed on certain chars, as noted in
Table 8, resulting in standard deviations in general of 1 to 2.5% of the mean,
except for redwood pyrolyzed at 15°C/min to 500°C for which the standard
deviation was much larger (14%).
TABLE 8. SUMMARY OF SURFACE AREA ANALYSES ON SELECTED LIGNOCELLULOSIC CHARS
Surface Area on Basis of
Material
Pecky Cedar
Redwood
Corn Stover
Computer Paper
*
°C/min
15
15
15
1
1
1
15
15
15
15
1
1
1
1
1
1
TF
°C
500
700
900
500
700
900
500
600
700
900
500
700
900
500
700
; 900
Chara
(m2/g)
380
486
539
449d
534d
650e
305e
463
489
500
296
394
465
353
421
385
Carbon
-------
o
Specific surface areas expressed as m per gram char, gram carbon, or
gram educt are presented as a function of final pyrolysis temperature in
Figures 8, 9, and 10, respectively. The shaded areas in these figures repre-
sent the surface area observed in Cellulose M chars pyrolyzed at heating rates
between 0.03°C/min and ll°C/min, as found by Brunner and Roberts (29).
Figure 8 indicates that for redwood lower heating rates result in higher
surface area for a given final temperature, much as was found for cellulose.
Comparing the redwood plots with the shaded area representing cellulose, it is
apparent that redwood heated at l°C/min has a surface area generally equal to
or greater than cellulose heated at 0.03°C/min. This indicates that there may
be a critical low heating rate below which little gain in surface area is
realized. Further work would be necessary to verify this, however.
As might be expected, Figure 8 shows lower surface areas on a total
weight basis for the higher ash materials, corn stover and computer paper.
This is at least partly explained by the high content of ash in the chars
(~ 25% in computer paper pyrolyzed to 900°C) which would contribute little to
the measured surface area (91). To isolate the surface area development in
the carbonaceous fraction of the chars, surface area based on carbon is dis-
played in Figure 9. When compared with Figure 8, the carbon-specific areas
(Figure 9) exhibit a narrower range of surface areas observed for the widely
varying lignocellulosics, especially for Tp = 700°C, at which surface areas
for all materials and heating rates studied fall within ± 10% of the median
value. Assuming that the standard deviations observed for surface areas of
redwood chars pyrolyzed at l°C/min apply to the pther materials, the plots of
surface area per gram carbon are statistically indistinguishable (95% confi-
dence intervals overlap).
The results of this study of surface development upon pyrolysis of ligno-
cellulosics indicate that, in general, surface area per gram of carbon varies
relatively little with heating rate or educt. Therefore, when compared on the
basis of surface area produced per gram of educt (Figure 10), the higher yield
materials (e.g., those with higher lignin contents) exhibit a substantially
higher production of surface area.
In conclusion it appears that at least on the basis of surface area as
determined by C02 adsorption at 195K, a wide range of lignocellulosics could
serve as starting materials for activated-carbon production, provided the ash
content of the chars or activated carbons could be reduced sufficiently.
Commercially this is accomplished, especially for ash reduction of the final
activated product, by washing with water and/or mineral acid solutions (1).
Also, ash reduction of chars prepared from municipal refuse has been shown to
be feasible by air classification (92).
ACTIVATION OF LIGNOCELLULOSICS AND REFUSE-DERIVED FUEL
Introduction
The purpose of this section of the study was to investigate the effect of
extent of activation on the porous structure of lignocellulosic chars, and
56
-------
< 500
cr
400
LU
O
or
300
200
100
V PECKY CEDAR
o REDWOOD
D CORN STOVER
O COMPUTER PAPER
CELLULOSE
°C / min
°C / min
A_ » i 1 i 1
0 * 500 700 900
FINAL TEMPERATURE(°C)
Figure 8. Surface area of char versus final temperature.
determined by CC^ adsorption at 195K.
Surface area
57
-------
^600
o
CM
UJ
cc
UJ
o
oc
ID
CO
500
400-
30O
200
100
V PECKY CEDAR
o REDWOOD
D CORN STOVER
O COMPUTER PAPER
CELLULOSE
min
500
900
FINAL TEMPERATURE(°C)
Figure 9. Surface area per gram carbon versus final temperature.
58
-------
?
2OG
CM
UJ
rr
^ 150
100
50
LLJ
O
oc
V PECKY CEDAR -
O REDWOOD
a CORN STOVER
O COMPUTER PAPER
CELLULOSE
i i- i
5OO
70O
900
FINAL TEMPERATURE (@C)
Figure 10. Surface area per gram educt versus final temperature.
59
-------
thereby to prepare activated carbons of lignocellulosic origin for comparison
to commercially available activated carbons on the basis of aqueous sorption
performance. Our approach to the investigation of activation was to select an
educt (prune pits) and activation conditions which would be expected to yield
an activated carbon with favorable sorptive properties, based on studies of
similar materials by Marsh et al. (2,3). The second phase of the investiga-
tion was to pyrolyze and activate RDF under the same conditions that yielded
desirable activated carbons from prune pits. The second phase was intended to
allow comparisons of the activation behavior of the very dissimilar educts
and, further, to clarify the potential for the production of activated carbon
from municipal waste.
Activation of Prune Pit Char
Prune pits which had a 0.62% ash content were first size-fractioned to
74-351 pm diameter (45 x 200 mesh) and pyrolyzed as described above. The
temperature in the furnace was raised from room temperature to 900°C at ap-
proximately 15°C/min. This resulted in a char yield of 25.02 ± 0.11%.
This char was then activated in a controlled C02 environment at 900°C for
times of 15, 30, 42, and 60 min (these will be designated henceforth as 15M,
30M, 42M, and 60M, respectively). Three to five samples were formed for each
of these activation times. The percent mass loss during activation increased
linearly with time of activation, as shown in Figure 11. Char activated for
15 min lost 21.65% ± 0.83% of its mass whereas char activated for 60 min lost
67.22% ± 2.69% of its mass. These values, along with N2~adsorption data, are
summarized in Table 9. '
The surface area and pore volume distribution of the activated prune pit
chars were determined by N2~adsorption and mercury porosimetry.
^-Adsorption Isotherms
The ^-adsorption isotherms were measured in duplicate for 60M, 42M, 30M,
and 15M (Figure 12). For comparison, N2-isotherms were also conducted on
three commercially available activated carbons. Filtrasorb 400, Filtrasorb
100, and Aquanuchar A (F400, FlOO, and AN-A, respectively). These results are
presented in Figure 13.
. as described by the
The volume in pores with radius less than a given
From the ^-isotherms and the BET equation (Eq. 1) was calculated surface
area. Pore radius corresponds to relative pressure P/Pc
Kelvin equation (Eq. 4).
value has been determined by the Cranston and Inkley method (12). The Kelvin
equation can correctly measure the radius of only those pores with radius
greater than 1 nm (7) (although it can incorporate the summation of pore
volumes with smaller radii). Furthermore, above P2/ps = 0-9 (which corre-
sponds to r = 10 nm), data are prone toward error and so cannot be quantita-
tively interpreted. This is because the value of P is .very dependent on the
temperature of the liquid nitrogen bath, which fluctuates slightly.
Included as ordinates in Figures 12 and 13 are both gaseous N, and liquid
N£ pore volumes. The gaseous N2 volume measured at STP within the adsorption
60
-------
100
0 20 40 60 80
ACTIVATION TIME (MINUTES)
Figure 11. Mass loss versus activation time for prune pit char.
61
-------
Si IW
P4 O
t>, o) ra
H I a.
3 s o a
o -H > os
u a
01 O 0) 43
S >4 VI 4J
, 0 S
>
13
H
3
CMC 2
5=>''-1
(DO
00>.
a co
a)
Pu
1
S3!
CO
in
co
vo
co
in
t>.
CM
C3
s
H
o
c3
o
1-i
_<->
g
d
o
M-l
d
0
H
4J
a
o
73
h
a
a
a
4J
CO
H
§
§
4J
d
01
CO
0)
u
04
01
K.
00
01
3
rH
a
^r
.
a
01
u
a
0)
o
a
U-l
H
I
(0
EH
pa
i
CM
01
43
4J
^
01
a
i-i
1
a
o
to
H
01
00
a
H
01
>
0
a
01
u
a
>4-l
H
3
5
01
a
1
§
0)
4J
a
I
a
H
4J
a
4J
H
4-1
d
§
cr
o
|
H
a
H
01
o
jj
g
H
4J
H
m
4-1
a
CO
d
o
4J
a
r-4
3
O
a
u
H
H
pa
o
p*
01
(0
01
n
o.
01
ra
01
O]
01
J3
4J
d
l-l
a
d
H
00
01
3
r-l
a
>
a
4J
H
pa
i
CM
43
4J
a
4J
a
a
EH
Ed
pa
I
g
01
2.
a
o
u
62
-------
550,
0.800
- 0.700
0.600
-0.500
0.40O
- 0.300
-0.250
o
.a
w-
o
o
8
o
o
CM
3
cr
P / P
r.2 S
Figure 12. Nitrogen isotherms for 60M, 42M, 30M, 15M, and F400. Filled in
and open symbols represent experiments 1 and 2, respectively.
63
-------
F400
F 100
o AQUANUCHAR A
- 0.600
o
e
o
o
T3
Q)
P/P
2 S
r- 0.500
O
>
TJ
- 0.400
o-
1
- 0.300
iXJ
Figure 13. Nitrogen isotherms for F400, FlOO, and AN-A. Filled-in and'opened
symbols represent experiments 1 and 2, respectively.
64
-------
equipment can be converted to the equivalent liquid No volume at 77K by multi-
plying the former by 0.001558. The liquid N, volume corresponds closely to
the true pore volume within the activated carbon. Pore volumes for r < 3
nm and r < 10 nm, along with surface areas, are shown in Table 9. No-
adsorption equilibrium times were generally about 10-30 min, although times of
90 min were required for the 15M carbon during the first several increments.
In the nitrogen-adsorption analysis of prune pit char (CHAR), however,
adsorption equilibrium was not reached within 72 hours. This suggests that
the diffusion of N~ into' the pores was extremely slow.
h°wever> could
enter these pores more rapidly; equilibrium was reached within 30 min for
adsorption at 195K. The difference in rate is due to a much higher activation
energy for diffusion of nitrogen at 77K than for CO, at 195K (13). Based on
C02 isotherms at 195K, the char's surface area was calculated by the BET
equation (Eq. 1) as 482 m /g. The surface area estimated from COo data at
195K should not be compared quantitatively to the estimate based on No~
adsorption.
, To test the reproducibility of the activation experiments, No-adsorption
analyses were performed in duplicate for each of the three samples which were
activated for 15 min. The percent variation in surface area should have been
greatest at the shortest activation time since the timing error introduced by
startup and shutdown would be the most significant under those conditions. If
physical characteristics such as percent burnoff and surface area varied lit-
tle among the samples activated for the shortest time, it would be reasonable
to assume that they would vary even less among those samples with longer acti-
vation times. As can be seen IA Table 10, variation in percent burnoff and
surface area was indeed slight among the 15-min carbons. These results justi-
fied the mixing of carbons prepared by activation for a given time period to
obtain adequate quantities for the aqueous sorption studies described later.
From the ^-adsorption isotherm data in Figures 12 and 13, it can be seen
that the volume of pores measurable by N« adsorption (r < 10 nm) increased as
activation time increased. Further, the corresponding values of the pore
TABLE 10. RELATION BETWEEN SURFACE AREA AND BURNOFF FOR CHARS
ACTIVATED FOR 15 MINUTES
Experiment
Number
1
2
3
Percent .
Burnoff
22.50
21.59
20.85
Surface Area
m /g
674.6 ± 0.4
660 ± 5
654 ± 8
Average
21.65 ± 0.83
663 ± 10
Note: Surface area values are mean ± standard deviation for duplicate
experiments on each sample.
65
-------
volume (r < 10 nm) of Filtrasorb 400 and Filtrasorb 100 were Intermediate
between those of 42M and 30M. Filtrasorb 400 had slightly more small-pore
volume than F100. AN-A had a small-pore volume similar to that of 30M. The
60M prune-pit-derived activated carbon had a small-pore volume more than 50%
greater than any of the commercial products.
Likewise, specific surface area increased substantially both with activa-
tion time and with consequent increase in percent mass loss. Whereas the
prune pit char activated for 15 min had a specific surface area of 663 ± 10
m /g, the prune pit char activated for 60 min had a specific surface area of
1692 ± 8 m/g, a value two-and-a-half times greater. Specific surface area is
plotted as a function of percent mass loss in Figure 14.
The No-adsorption data in Table 9 reveals that the volume of monolayer
surface coverage was a fairly consistent fraction (0.8) of the total pore
volume with radius less than 3 nm. .The volume of monolayer surface coverage
(V ) is used to calculate surface area (S), as described previously (Section
4). Including the appropriate constants for N, in Eq. 3 gives the following:
S(m2/g)
2792 Vm (cnrYg)
(17)
Mercury Porosimetry
Mercury porosimetry measurements, which describe the macro- and
transitionalpores of a porous solid, were conducted fon CHAR, 15M, 30M, 42M,
60M, F400, F100, and AN-A. The initial results for all of the carbons ana- ,
lyzed were of an S-shape similar to that for 60M in Figure 15. (The results
for the other carbons are in Appendix B.) As can be seen in all of these
curves, there is a great increase in penetration volume corresponding to a
pore radius between 15,000 and 4,000 nm (or pressure of 8 to 20 psi). This
increase is an artifact; it corresponds to the pore volume between the sample
grains rather than within them and is not useful information for our purposes.
Can et al. (91) found that pressures below 60 psi (r = 1500 nm) corresponded
to intraparticle voids for a 40 x 70 mesh fraction of both crushed glass arid
non-porous coals. Since the carbons we used were a factor of five smaller (200
x 400 mesh), it can be expected that the interparticle voids, which are as-
sumed proportional in size to particle diameter, would occur down to a void
radius of 300 nm (corresponding to 300 psi). This, then, was our assumed
dividing line between intra- and interparticle porosity.
Indeed, a comparison of all of these penetration curves for the 200 x 400
mesh size reveals that if the penetration volume corresponding to 300 nm (300
psi) is subtracted from the penetration volumes corresponding to smaller pore
radii, the data from the prune pit carbons appear reasonable, as shown in Fig-
ure 16, and the intraparticles voids no longer appear to be a significant
factor.
The adjusted values of mercury penetration increase regularly with in-
creasing activation time. The 15M activated carbon differs from the char only
in the small transitional- and micropore range. The 30M, 42M, and 60M chars
66
-------
o»
CM
UJ
i2
CO
CO
Figure 14.
1800
1600
^ 1400
2
CC
< 1200
1000
S£ 800
o
600
UJ
CO 4OO
I
200
0
T r
0
20
40
60
80
100
PERCENT MASS LOSS
specific surface area versus percent mass loss for acti-
vated prune pit chars. Error bars are included to represent
standard deviations when the standard deviation is greater than
the magnitude of the corresponding symbol.
67
-------
o
4J
(U .H
a x
0 0)
O -H
m ra
o
M V-i
0) O
4-> ft
o
13 to
-a
o) a
P O
O ft
ft W
Q)
to $-1
3 VJ
CO O
n o
gj
> -H
CO
0) ft
§0
rH O
O CO
a
O iH
H 0)
4-1 ^2
ca
P CO
4J CD -^N
cu M oo
> ft
H VO
4-1 VJ
c8 O II
i-H IM
3 -H
S d to
3 O ft
O -H
4-J i-l
CO «-'
M
>. 4J
!-l 0)
4-1 (3 CO
O> QJ C
B ft -rl
H CO
CO pi ^
O -H «)
&§
M CO
3 CU CJ
O W)
M V4 CU
CU CO £1
68
-------
a.
NOLLVdl3N3d
3AI1V13H
CO
S
w
J-i
a /-N
o cd
us
, nt
!-i t^
4-1 OO
<1J
(3 \O
0)
ft II
i-t CO
3 04
o
O
cu
? 8
a) «
a:
CO
3
O
rl VO
4J
n) vi
cd
4-1 J-l
0)
E cu
H H
CO O
O ft
M
aj-l
a
a)
M cd
S >
CJ vH
M 0
cu cr
g cu
vO
,-1
cu
69
-------
exhibit successively larger pore volumes in all size ranges, with the greatest
increases occurring in the size range less than 10 nm.
The mercury porosimetry data for F400, FlOO, and AN-A, shown in Figure
17, reveal that F400 and FlOO have much the same macro- and transitional-pore
structure. The pore volume of AN-A is considerably less than those of F400
and FlOO.
Pore Volume Distribution
Total pore volume including micro-, transitional-, and macropores was
estimated for each of the activated carbons. This was achieved by summing the
N2~adsorption and mercury porosimetry data; the total pore volumes are shown
in Figures 18 and 19. The N2~adsorption data were used to determine pore
volume up to a radius of 10 nm, and the mercury porosity data were used for
pore volumes with a radius larger than this. The mercury porosimeter achieved
pressures up to 60,000 psi, which corresponds to a pore radius of 1.5 nm.
Ideally, then, the pore volume distributions as determined by mercury porosim-
etry should be equal to those calculated by N2~adsorption over the range for
which both methods can be used for estimation, namely 1.5 nm to 10 nm (10 nm
corresponds to a pressure of 8800 psi). Such a comparison, tabulated in Table
11, shows that pore volumes calculated by mercury p'orosimetry were not signif-
icantly higher than those calculated by N2-adsorption at these pore radii.
This is somewhat surprising in light of the structural alteration that mercury
porosimetry is often thought to cause in the analysis of carbonaceous materi-
als (Section 4). Our results indicate that such alteration, which would have
yielded artificially high values for pore volume, may not have occurred in the
porosimetry analysis of prune pit char. However, the pore volume determined
by N2~adsorption was considered to be the more accurate of the two methods for
the region of pore radii less than 10 nm.
The data in Figure 18 show that the total pore volume with r < 300 nm
increased with increased activation time and percent burnoff. Whereas the 60M
carbon had a total pore volume of 0.930 cm /g, the 15M carbon had a total pore
volume of only 0.363 cm /g. Figure 18 does not include char values because,
as described above, N2-adsorption measurements of the char were not feasible.
TABLE 11. COMPARISON OF PORE VOLUMES DETERMINED BY N2-ISOTHERMS AND
MERCURY POROSIMETRY FOR 60M, 42M, 30M, 15M, AND F400
Pore Volume (cm3/g)
Pore Radius
(1 < r < u)
1.5 < r < 3
3 < r < 10
60M 42M 30M 15M F400
N2 M.P. N2 M.P. N2 M.P. N2 M.P. N2 M.P.
0.070 0.080 0.044 0.040 0.015 0.010 0.007 0.020 0.022 0.070
0.053 0.060 0.030 0.030 0.029 0.023 0.011 0.015 0.047 0.050
70
-------
CO
CO
M
8!
C
o
H
4-1
ni
(1)
0
(U
o
cy
T3
O) £
« o
t-H O
s .
O
0) O
4J
ni
CO
-H
4J H
-H
3AllV13d
a
H
ft)
71
-------
cu cu
8
ro
O
O
CM
O
8 -
W
CM
^ CO
§
10 {£
CVJ
2
o cu »-i
M 4-> O
cu cu ft
CO i-J
CU S
co cct cu
a
O CO *^J
CO CU CU
H M e
a cu 'a
o P 5
H CU
4-1 CO 4J
ft r-l CU
w I
k- *
Q)
"o>
o
&
a
>>
I C cfl
CM cu >
is a.
T3 O CD
M H ft
CU O I-l
> M-l O
r-l CU ft
p e cu
> -H W)
CO O
CU O !-i
O O -H
oo
.60
H
72
-------
(6/£uuo)3iAiniOA 3dOd 3AllViniAinO
i
VJ -rl
3 CO
.CJ >> O
J-i ,0 !-(
a) o
01
M CJ
CO 0) 5-1
4-1
O
tfl
o S
H 4-1 1-1
U C 0)
ft CU 4-1
M CO CU
O Cl) T3
CO !-J
TS ft CO
cd cu cu
CM rH
!2i CO tfl
rH >
e o
O ifi 4-1
n e a
m Fi cu
01 CO
T3 CU
CU p! 5-1
PI ^) Pi
H ft CU
eon
0) ,
03
co
O C
01 -* o
3 Pn -H
CO -U
M H ft
0) O P
> <4-l O
01
0) >~,T3-
g H CD
3 4J I
iH 01 S
O S CD
> -H txO
CO O
cu o M
M n -P
o o -ri
PM ft C
S,
H
73
-------
The F400 pore volume distribution was very similar to that of the 42M
prune pit carbon; F100 had a micropore volume slightly less than F400 but
otherwise similar. AN-A had a pore volume distribution between that of the
15M and 30M carbons.
The percent of the total pore volume that was of relatively small pore
size (r < 3 nm) is shown in Table 9, along with various other pore volume
data. The percent of pore volume within this arbitrarily chosen range (r < 3
nm) remained fairly constant at 73-83% (Table 9). Fundamental considerations
would suggest that in the initial stage of gasification, formerly closed
micropore volumes would be opened as key atoms blocking entrance to these
pores are gasified. At the same time, the walls of the already opened pores
would be gasified, causing an increase in the pore radius of any given pore.
Initially, the effect of this process would be to increase both the specific
surface area and the total pore volume (4). Moreover, if the volume of pores
opened by gasification were greater than the volume increase from wall gasifi-
cation, then the percent of total pore volume in the micropore range would
increase. These exposed key atoms may gasify more readily than those along
the wall surface because the exposed atoms probably have fewer and weaker
bonds holding them in place (31).
Eventually, however, it is expected that the reservoir of closed pores
would run out. At this point only those pores which had already been opened
would be gasified. This would cause a net increase in the radius of all
pores. At a further extent of gasification, the walls between pores would be
completely gasified away, causing various formerly separated pores to merge
into one large pore (4). The net effect of this process would be an eventual
decrease in the specific surface area for a continued rise in total pore
volume but decrease in percentage of pore volume within the micropore range.
For the gasification of the prune pit chars, however, such a decrease in
surface area and percent small pore volume was not observed, even after 67%
burnoff of the material.
Activation of Refuse-Derived Fuel
Refuse-derived fuel (RDF) char was prepared by pyrolysis under argon at
approximately 15°C/min to a final temperature of 900°C. The RDF initially had
an ash content of 11.68%; the complete composition was described earlier. The
char yield for these pyrolysis conditions was 30.6% on a gross weight basis
(standard deviation = 0.2%); the ash-free char yield, assuming no loss of
inorganics, was 21.4%; the yield of carbon was calculated as 44.0%. As expec-
ted these results indicate a somewhat greater loss of carbonaceous material
than was observed for prune pits pyrolyzed under the same conditions (ash-free
char yield equal to 24.6%, carbon yield not determined). This expectation is
based on the fact that, compared on the basis of organic material present in
the educt, prune pits have a higher lignin content than RDF (approximately
35.7 versus 17.8%).
The RDF char was activated in a flow of C02 at 900°C. Due to the caking
tendency of the char, a high C02 flow rate was necessary to effect even mar-
ginally homogeneous fluidization during activation. The flow rate (~ 3 1/min)
74
-------
was significantly higher than that used in the activation of the prune pit
char (~ 580 ml/min). To allow comparisons, activation of the RDF char was
conducted for periods of time which resulted in burnoff (mass loss) based on
carbon similar, to the 42M and 60M prune pit activation experiments. Table 12
presents a summary of the results of the RDF activation, derived from Table
9. Table 12 also presents the results of surface area determinations of the
activated carbons, using the BET equation (Eq. 1) to interpret N~ adsorption
at 77K. Since RDF char and activated carbons had such high ash contents, burn-
off (mass loss) and surface area based on carbon are also presented to allow
meaningful comparison with the prune pit results. Percent carbon values of
the prune pit char and activated carbons were not determined. However, since
the ash content of prune pits is on the same order as pecky cedar and redwood
(i.e., < 1%), the carbon content of the char is reasonably estimated as simi-
lar to that of those materials pyrolyzed under similar conditions (i.e., 96.8%
to 98.7% from Table 12). Thus the conversion of burnoff or surface area from
gross weight basis to carbon basis would be small for the prune pit char and
low-burnoff activated carbons (perhaps 5% higher surface area values, for
example).
It is evident from Table 12 that mass loss for the RDF char occurs more
rapidly than the prune pit char (approximately 65% carbon burnoff in 15 min
versus 60 min). However, much more burnoff is required to produce a given ,
surface area in the RDF char; e.g., 65% carbon burnoff of the RDF char yields
a surface of approximately 930 m /g carbon whereas only 37% burnoff of the
prune pit char achieves the same specific surface per gram carbon. The RDF
TABLE 12. COMPARISON OF ACTIVATION OF PRUNE PIT CHAR AND RDF CHAR
Mass Loss (%)
Surface Areac
Time of Carbon -
Activation Content
Material (min) (%)
Gross Wt. Carbon
Basis
Basis
D
m2/g m2/g-Carbon0
Prune Pit Char
RDF
11
" . '
11
Char
"
15
30
42
60
9
15
ndd
nd
nd
nd
49.1
39.6
21.7
37.1
49.7
67.2
30.3
42.8
_e
-
-
-
47.5
65.2
663
927
1175
1692
300
370
-
.
-
612
934
aDetermined using BET equation on N2 adsorption data (77K).
K, n , . , , . . irir. (Carbon content of activated char) , A
Tflass loss (carbon basis) = 100 - 77; -r (100 -
(Carbon content of char)
Gross Mass Loss).
o o
Surface area (m /g C) = surface area (m /g)/carbon content of activated char.
nd = not determined.
e(-) = cannot be calculated since percent carbon data not determined.
75
-------
char loses mass more rapidly during activation with less development of sur-
face area per unit mass lost. This comparison, however, is not general since
the conditions of activation of the two chars are quite different (much higher
C02 flow rate for RDF char activation).
In fact the different behavior of the RDF char in activation can be at
least partly explained by the higher C02 flow rate and the much higher inor-
ganic content of the char. It is thought that low flow rates of CC^ allow
retention of the gasification product CO at the char particle surface, result-
ing in inhibition of gasification of the surface and, therefore, greater
development of microporosity per unit mass loss (50). Such was the case for
prune pit char activation, which resulted in extensive microporosity and sur-
face area development. The higher C^ flow rate in RDF char activation may
have resulted in a higher percent of the observed mass loss occurring at the
particle exterior at the expense of development of microporosity and therefore
surface area. Furthermore, inorganic impurities are known to increase the
rate of gasification (51) and are thought to concentrate on the particle sur-
face (51), resulting in preferential development of macro- and transitional-
pores (53). Hence, even for equivalent C^ flow conditions, RDF char would be
expected to yield less microporosity and surface area per unit mass loss than
prune pits or other low-ash materials.
These results are consistent with past work on RDF pyrolysis and activa-
tion (92), in which activated carbon made from RDF was found to have a large
proportion of pore volume in the transitional-pore size range. In these stu-
dies the activated carbon prepared from RDF was found to have a low specific
suface area (350 m /g), similar to our results. The encouraging aspect about
this past work, however, was that the waste-derived activated carbon, despite
its low specific surface area, was equally effective as commercial activated
carbon (AN-A) in reduction of organics (measured by chemical oxygen demand) in
primary-treated municipal wastewater.
In conclusion our results and past work indicate that an activated carbon
can be prepared from classified municipal solid waste (RDF) which, despite a
low specific surface area and high ash content, may be a useful adsorbent for
some applications such as wastewater treatment. Ash removal by washing (1),
air-classification (92), or both might yield an activated carbon of generally
acceptable quality. Finally, preparation of the char by pyrolysis to 900°C
followed by activation with CO 2 at 900°C and lower flow rates than used in
this study may result in an even more attractive activated carbon with more
highly developed microporosity and surface area.
SORPTIVE PROPERTIES OF ACTIVATED LIGNOCELLULOSICS
Experiments were conducted both to determine the rate of approach to
equilibrium for the various carbons and to determine an adsorption isotherm
for the prune pit carbons and for F400. All DOC adsorption experiments were
conducted with filter-sterilized unchlorinated secondary effluent from the
Palo Alto Water Quality Control Plant.
76
-------
Rate Experiments
F400, F100, and AN-A
An experiment was conducted to compare F400, F100 and AN-A with respect
to their rates of uptake of DOC from secondary effluent. In this experiment,
a dose of 50 mg activated carbon/115 ml wastewater was used. Wastewater
included 100 ml of secondary effluent plus 15 ml of Milli-Q water in which the
activated carbon was presoaked overnight. The pH did not change significantly
in this experiment, remaining in the range 7.5 to 8.0.
The kinetic behavior of Filtrasorb 400 and 100 were much alike in this
experiment, as shown in Figure 20. Of an initial DOC concentration of 14.2
mg/1, only 9 mg/1 remained after 20 min and 6.3 mg/1 after 120 min. The AN-A
adsorbed less rapidly; the residual concentration after 120 min was 8 mg/1.
Prune Pit Carbons and Filtrasorb 400
Two adsorption experiments were conducted on 60M, 42M, 30M, and F400,
with a dosage of 30 mg carbon/110 ml wastewater (runs 4 and 11). Coinciden-
tally, the two secondary effluents used for these two experiments had vir-
tually the same DOC of 10.04 ± 0.60 and 10.20 + 0.55 mg/1 for runs 4 and 11,
respectively. In light of this similarity, the results of these two runs were
averaged, and a 95% confidence interval was established for the average. In
establishing a 95% confidence interval from two data sets, one must assume
that the two data sets represent subgroups of the same population. Statisti-
cally, such a claim could not. rigorously be made in this case; although the
DOC of filtered secondary effluent on one day may by coincidence be the same
as on another day, both the compounds which constitute this DOC and the prop-
erties of these compounds for the same two samples may vary considerably.
Such problems are inherent when the complexity of secondary effluent is ig-
nored by using a collective parameter such as DOC. The data from runs 4 and
11 are shown in Appendix C. The average of the two runs are in Figure 21.
The 95% confidence interval for these averaged values are generally of a,
magnitude corresponding to the size of the symbols that are used tofrepresent
the data.
In run 4, the magnitude of adsorption at any given time increased with
the magnitude of surface area. Adsorption is greatest for 60M, followed in
order by 42M, F400, and 30M. In run 11, this same trend persisted with the
exception that F400 and 42M behaved nearly the same. Moreover, for F400 the
variation between the adsorption behavior of run 4 compared to run 11 was
slightly greater than the variation experienced by any of the other activated
carbons, especially at the longer times. For most of the carbons the varia-
tion in DOC at any given time, from run 4 to run 11 was on the order 0.5 to
1.0 mg/1. In comparison, the standard deviation varied from 0.10 to 0.65 mg/1
for triplicate analyses for a given set of conditions.
Also included in run 11 are several data for 15M and Char at prolonged
times, as can be seen in Appendix C. The 15M adsorbed less than did 30M, and
the Char adsorbed less than did the 15M.
77
-------
LU
81
H
O
in
60
d
cd
o
4-1
cd
CN
C
t-l
O
O
O
O
o
H M
+J 0)
ft 4J
!-i t8
O !5
CO 0)
T3 4-1
Cti 0}
o o
I I
OJ r-1
4-J O
0) rd
CO
O
(l/Buu) 9NIN!VIAI3d OOQ
O
-------
CO
o:
ID
o
LU
CO
CU S^ rH
> m o .
u a
O 60 >»
cd ft co
H
60 4-1 CU
& fi £*
CU 4J
O CO
co cu IH
VJ O
o cu cu
H TJ
60 3 .
C CO 4-1 d
H J-l -H O
-a cd c -H
co ,a 60 co
i-r j-i S t-i
O C
CO V) CU O
J-l JS O
4J W fJ
CO T3
C -H
O ^N T-l O
^5 * 1^ ^
o -5
o
4J CU
pt ] 1
!-i cd
O 3
CO CU
cd co
cd
u &
O
*§
C co
CU CO
ft cu
C CO
J3 cu
9 S
4J
O
cu
4-1
O O
9NI.NIVW3d OOQ
CO
o
H a
4-1 O
CU ,£>
H cd
J*J CJ
CN
cu
n
60
H
CO 4-1
> cu
1-1 rH
CU CU
4J 13
a
H CU
a) cd
t)
r-l C CO
H rQ
-------
Adsorption Isotherms
Adsorption isotherms were conducted on 60M, 42M, 30M, and F400; the
results of these experiments are plotted as q (mg DOC adsorbed/g carbon)
versus C& (mg DOC remaining/1) in Figure 22.
The development of adsorption isotherms required that the ratio of ini-
tial adsorbate (grams) to activated carbon (grams) be varied over a range of
several orders of magnitude. In principle, this can be achieved by one of
several methods. The first entails varying the concentration of the initial
adsorbate over several orders of magnitude for a constant water volume and
constant activated carbon mass. Alternatively, for a constant initial concen-
tration of adsorbate, the ratio of activated carbon mass to water volume can
be varied. The latter of these two is usually used, and was considerably more
practical in our case in which the adsorbate consisted of the organic sub-
stance present in secondary effluent. It would not have been feasible to vary
the concentration of residual organics over a wide range.
For these reasons, the ratio of activated carbon mass to wastewater vol-
ume was varied; the initial DOC concentration was maintained constant. In the
experiments the carbon/wastewater ratio was varied from 0.990 g/25 ml to 0.015
g/200 ml, which represents a range of a factor of 500. It was found that the
standard deviation in q (calculated as described below) was too great for
carbon/wastewater ratios below the lower extreme of 0.015 g/200 ml. Although
several experiments were conducted with lower ratios, the results were not
statistically useful.
From the results of these experiments (designated as run 8), it can be
seen in Figure 22 that 60M adsorbed the most DOC, followed closely by F400.
Trailing behind these are 42M and then 30M. For the prune pit carbons, then,
the magnitude of adsorption increased with increased activation.
The standard deviations for the data also are shown in Figure 22.
standard deviation of C
given sample; that of q
The
e was taken to be that of the triplicate analyses for a
was calculated as
(18)
where GC ± Sc = the mean DOC concentration of the control ± its standard
deviation; Ce ± Sc = mean DOC concentration of the given sample at equilib-
rium ± its standarl deviation; V ± ST7 = volume of wastewater ± an expected
v
experimental standard deviation taken to be 2%, and M ± S was the mass of
the sample ± the experimental standard deviation inherent in weighing. As can
be seen in Figure 22, the standard deviation became great as C approached
C . The value q was calculated as
q. (mg/g)
(Cc - Ce) (mg/1) x y (1)
(19)
80
-------
o
CD
(T
O
Q
UJ
O
UJ
CD
'00
80
60
§40
O
O
Q
E
^ 20
-------
q rather than the
The control concentration C was used in the calculation of
initial concentration C . The value C represents the DOC remaining in a
controlled reaction flask which was subjected to the same conditions as was
used for the samples.
carbon present.
In the control flask, however, there was no activated
It was the control concentration, rather than the initial concentration,
with which adsorbed and bulk phases came into equilibrium.
Interestingly, linear isotherms most accurately fit the adsorption behav-
ior of each of the activated carbons. Correlations with both Freundlich and
linear isotherm relationships were investigated for these data; these models
were not as successful in describing the data as were a linear isotherm. A
plot with Langmuir coordinates (l/qe versus l/Ce) showed very little detect-
able pattern for any given carbon.
The Freundlich isotherm model fit all but the lower ranges of log qe
versus log C data. Experimental fluctuations in the data at these low ranges
were exaggerated by the log-log type of a plot. The Freundlich equation is of
the form:
l/n
(9)
The values of n calculated from these data were generally close to 1 and a 95%
confidence interval encompassed 1, as shown in Appendix D. This suggests that
a linear isotherm could be as easily used to describe the data.
The linear isotherm was of the form:
- Cn>
(10)
where C corresponded to the non-adsorbable portion of the DOC. The linear
adsorption constants were determined by linear regression of the data in
Figure 22. These constants, as well as the 95% confidence intervals are shown
in Table 13.
TABLE 13. LINEAR ADSORPTION ISOTHERM COEFFICIENTS
Activated
Run No.
Run 8
Carbon
Type
60M
42M
30M
F400
Calculated
Nonadsorbable3
DOC C
(mg/1)
1.28
1.15
-0.44
0.95
95% Confidence
Interval of
b = qe/(ce~cn)
(liters/g)
17.5 ± 7.3
6.5 ± 2.0
2.5 ± 1.4
11.4 ± 2.0
Correlation
Coefficient
(r)
0.9340
0.9549
0.8683
0.9763
Number
of
Points
7
8
8
10
Run 6
F400
0.97
5.4 ± 1.4
0.9333
13
Defined as the intercept on the abscissa of Figure 22.
82
-------
The 95% confidence interval of b was computed as (93):
191 1/2
b± t(0.975, n-1) [(f-y) bZ(-f - l)]
XI £f £,
(20)
where t is the t distribution, n is the number of data points, and r is the
correlation coefficient,
The nonadsorbable concentration in these linear regressions was fo^. most
cases calculated to be about 1.0 mg/1. Indeed, in all the data collected in
run 8, an equilibrium concentration of DOC never was achieved much below 1.0
mg/1, even at very high dosages of activated carbon. This fraction, then,
could be considered to be the non-adsorbable portion of the DOC; it consti-
tuted about 10% of the total control concentration.
A second adsorption isotherm experiment was conducted with F400 (run 6);
the results of this experiment are shown in Figure E-l (Appendix E) along with
the results of F400 in run 8 (the comparison isotherm experiment described
above), and the adsorption isotherm data for F400 from the equilibrium condi-
tions achieved in the kinetic experiments. As can be seen, the slope of the
curve for run 6 is less steep than for run 8. Correspondingly, the control
concentration for run 6 was nearly 2 mg/1 higher (11.78 mg/1) than for run 8
(10.09 mg/1).
The isotherm data from the kinetic experiments agreed closely with the
linear regression on the equilibrium isotherm results from run 8. The control
concentrations in these.kinetic experiments are shown in Appendix E, Fig. E-l.
The relationships between T (mg DOC adsorbed/nT specific surface area)
and Ce was also investigated, and is presented for run 8 in Figure 23. As can
be seen, F400 and 60M behaved much the same (within the statistical signifi-
cance of the data). 42M adsorbed less organic carbon per N^-BET specific
surface area at a given equilibrium concentration Ce, and 30M still less.
This would suggest that less of the specific surface area for 42M and 30M than
for F400 and 60M was available for adsorbing some portion of organic com-
pounds. The difference between the prune pit carbons was perhaps due to the
size of the pores; the small pores may have excluded some of the larger or-
ganic molecules from reaching potential adsorption sites to a greater extent
in the 30M carbon than in the 60M carbon.
The difference may also be due to variations in the functional groups
which are present in the activated carbon at its surface. This question was
not investigated in the present study.
Model for the Kinetics of Adsorption - . '
The equation for diffusion into a spherical adsorbent grain for which a
linear isotherm applies is of the form:
3C
3t
D_
(1 +
3r
2
r
(13)
83
-------
T3
0>
o
O
O
Q
0.09
0.08
007
0.06
o>
o
^0.05
=3
(O
§0.04
.a
o
o
0.03
0.02
0.01
0
o 60M
D42M
A30M
V F 400
0
246
Cp( mg/l )
8
10
12
*>
Figure 23. T (mg DOC adsorbed/m surface area) versus Cg (mg/l DOC remaining
in solution at equilibrium) for 60M, 42M, 30M, and F400 (run 8).
84
-------
This was described in Section 5 of this report. For the case where the carbon
is initially free of solute, and the bulk concentration decreases with time,
this expression is numerically solved in the form given by Crank (68):
where
a
T
where
Co *
- f = 1
6a(ot
-xq
n
L 2 ?
n=l 9 + 9a + q a
= initial bulk solution concentration, (g/m ),
o
= concentration of bulk solution at time t, (g/m ),
o
= concentration of bulk solution at equilibrium, (g/m ),
= fractional approach to equilibrium,
= non-zero roots of tan q
= (CQ - C00)/(Co - Cn) = equilibrium fractional uptake,
= (1 - F)/F,
= dimensionless time parameter given by
Dt
3q /(3+aq ) (Appendix F),
T =
(1 + R)a
(21)
(22)
t = time, (s),
o
D = diffusion coefficient, (m /s),
a = average grain diameter, (m), and
R = partition coefficient between the adsorbed and bulk concentration
of solute in a given activated carbon pore (dimensionless).
The diffusion equation can also be solved in another form that is more
convenient for evaluation at small values of the time parameter (at small
times the solution to Eq. 23 yields many roots). The alternative solution is
of the form:
f = (1 + a) [1 -
e erfcj-
85
(23)
-------
In this notation:
I =| {(1 + 4/3cc)/2+
(23a)
Y2
(23b)
e erfc z = exp z erfc z
Both of these expressions were programmed on a Hewlett-Packard 97 calcu-
lator to be used in modeling the kinetic behavior of the DOC.
Several assumptions are 'required for the use of Eq. 13. The first of
these is that diffusion occurs only in the bulk solution of the activated
carbon pores, and that the portion of solute that is adsorbed is not free to
diffuse along the surface of the pore. Moreover, it is assumed that the rate
of adsorption and desorption is much faster than the rate of diffusion, so
equilibrium between the pore bulk and adsorbed phase is considered to exist at
all locations within the pores. The assumption of pore phase equilibrium is
usually accepted in the literature (13,61,67,70). However, some of the liter-
ature indicates that surface diffusion does indeed occur (13,61,70) and often
contributes more to overall mass flux within the pores than does bulk diffu-
sion. In our experiments, using the collective parameter DOC, it was not
possible to discriminate between surface diffusion and bulk diffusion, because
the two cases are virtually identical mathematically when the equilibrium
isotherm is linear.
The second group of assumptions relates to the calculation of F, the
equilibrium fractional uptake. The quantity F equals the fraction of the
total solute initially present that was eventually adsorbed by the activated
carbon under the conditions of the experiment. It must be considered that the
non-adsorbable portion of DOC, C , does not contribute to a diffusion driving
force, and so should be subtracted from all values of concentration. The
value F, then, is calculated as
F =
([C - C ]- [C - C 1) C - C
*" o nj L °° nj o °
1C - C I
L o nj
C - C
o n
(24)
The third group of assumptions relate to the calculation of R, the parti-
tion coefficient:
R
r g solute adsorbed within the grain -i
'-g solute in bulk solution within the grain-'
R
rg solute adsorbed g sorbent^
g sorbent
3 "4
cm grain
C
g solute in pore solution cm pore volume-j
cm pore volume
cm graxn
86
-------
[b x 1000] x p
T> _
(25)
where
Pb
the slope of the linear adsorption isotherm determined from run 8,
(mg solute/g carbon)/(mg solute/1 solution)
o 3
(grams active carbon/(cm carbon volume + cm void volume)
o
(cm pore volume in the macro- and transitional-pore size
range)/(cm total pore volume).
It is assumed that the interior void volume of the activated carbon can
be determined as the summation of volumes calculated by N^ isotherm and mer-
cury porosimetry with radius < 300 nm. The bulk pore volume, representing the
pore volume that contains a bulk solution phase (as opposed to only an ad-
sorbed phase) is taken to be the interior pore volume which has a radius
> 3 nm. It was further assumed that the true carbon density (excluding voids)
was 2.1 g/cm .
e , and R, based on these assumptions, are shown in
and 'F400. The quantity of solute adsorbed, if
The values for b, p
Table 14 for 60M, 42M, 30M,
these assumptions are correct, is more than four orders of magnitude greater
than the quantity of solute in equilibrium with this in the bulk phase, as
indicated by the values of R in Table 14.
Based on these assumptions, the averaged data from runs 4 and 11 were
evaluated to see whether they exhibited behavior similar to that predicted by
the model. Theoretical plots of log f versus log T could be made through the
use of the two programmed numerical solutions (HP-97 calculator). Such plots
were made for 60M, 42M, 30M, and F400 based on the fractional uptake (F) that
was observed in each experiment. Of the two numerical solutions for the dif-
fusion equation, the second, involving the error function, was the most conve-
nient. In the first solution, the infinite series was truncated to six terms
corresponding to the six roots for q that were provided and are shown here in
TABLE 14. ESTIMATION OF THE PARTITION PARAMETER R
Active
Carbon
Type,
60M
42M
30M
F400
b
rmg/g^
Lg/lJ
17.46
6.48
2.53
11.40
p
3
f 8 W cm
I 3Jl
cm
0.711
0.900
1.010
0.895
et
pores with 3 nm < r< 300 nm->
cm total volume
0.131
0.1161
0.139
0.180
R
r g adsorbed -v
^g in solution'
94,900
50,200
18,300
56,700
87
-------
Appendix G (68). If the last of these terms is insignificantly small, then
the model is accurate; however as T becomes small, this last term becomes more
significant. Likewise, the terms which were truncated out of the infinite
series are also significant, and their exclusion from the infinite series
causes the numerical solution to become inaccurate. This significant error
occurs for values of T less than 1 x 10 , corresponding to a time of 20 min
in our experiments.
The second of the numerical solutions (Eq. 23), on the other hand, was
useful throughout the range of times. Furthermore, it agreed with the first
numerical solution to three significant digits for T greater than 10 .
From the HP-97 programs were determined the diffusion coefficient D which
resulted in the best fit between the experimental data and the theoretical
model at any given time. These calculated diffusion coefficients were fairly
consistent (within one order of magnitude, between 4 x 10 and 8 x 10
cm /sec) for all of the carbons studied. These are shown, along with T and
both the experimental f and calculated f, in Appendix G. The magnitude of
diffusion coefficients for 60M were slightly larger than the others. Then, in
descending order of D values came F400, 42M, and 30M. The difference between
these, however, is not significant, especially in light of possible errors in
estimating the several variables which were used to calculate the values of D.
For a given carbon, the diffusion coefficient was generally slightly
higher at the early time of 1 min, and slightly lower at the latter times of
60 min and 210 min. This is consistent with what might be expected: at very
short times, those compounds which diffuse the most rapidly tend to dominate.
As time passes, these rapidly diffusing compounds will have already reached
equilibrium, and those which diffuse more slowly will still be entering the
carbon. The prominence of these slower molecules would tend to diminish the
measured diffusion coefficient.
In a second approach to comparing the model to the experimental data, a
median diffusion coefficient was chosen from the range of D values determined
above for each of the carbons. This D usually corresponded to that measured
at 5, 10, or 20 min. The value so chosen was designated the effective diffu-
sivity Dg for the given experiment. Using this value D£, corresponding values
of T were calculated. The plot of experimental values of f and T (Figure 24)
shows good agreement between the experimental data (represented by symbols)
and the model (represented by the series of curves). These effective diffu-
sion coefficients are tabulated in Table 15, along with the corresponding
transitionaland macropore volumes.
These diffusion coefficients are compared (Table 16) with those predicted
for bulk diffusion of high molecular weight compounds and various specific
compounds (Reid and Sherwood (94)), and the pore and surface diffusion coeffi-
cients determined by several other investigators (61,70,79). As can be seen,
the pore diffusion model and experimental data described in this report were
successful in determining pore diffusion coefficients consistent with those
expected for large molecules. Assuming a median molecular weight of 500 for
organic constituents represented by DOC in secondary effluent (73,74,75). the
expected value of the bulk diffusivity for DOC is on the order of 3x10- m/s.
88
-------
o
1.0
0.8
0.6
0.5
0.4
o8
I 0:3
O
O
0.2
0.15
0.10
-F=0.842I
42 M
F400
FO.8476
o 60M, De= 2.9x10 m2/sec
a 42M , De= I .6xlO~'° m2/sec
* 30M, De=I.OxlO~lom2/sec
v F400, De=l.3xlO~' m2/sec
nil
I Mill
I i i I i i nl
I I
0.0001
0.001
0.01
Det
O.I
R)a
Figure 24. Pore diffusion model for kinetics of DOC adsorption by 60M, 42M,;
30M, and F400. Values of D£ corresponding to best fit (runs 4 and
11). Curves represent the solution to pore diffusion model as f
versus T for the appropriate value of F, the total fractional
uptake of DOC.
89
-------
TABLE 15. MEDIAN DIFFUSION COEFFICIENT AND PORE VOLUME
Active Carbon
Type
Average Effective
Diffusivity (nT/g)
Pore Volume
(cm /g) with
3 run < r < 300 nm
60M
42M
30M
F400
3
1
1
1
.2
.6
.1
,4
X
X
X
X
io-10
io-10
io-10
ID'10
0
0
0
0
.128
.105
.104
.160
The experimental values were in the range of 1 x 10~ to 3 x 10 m /s.
Values of D slightly lower than the molecular diffusivity are justified by
deviation o? the pores from the ideal model of straight, cylindrical tubes
(69). Frequently this deviation is expressed as
De = D/X
(26)
where
2
D = effective diffusivity for internal pore transport, m /s,
2
D s molecular diffusivity, m /s, and
X = tortuosity factor, dimensionless, accounting for irregularities of
pore shape.
Values of X have been reported in the range of 2 to 10 for porous solids such
as catalyst grains (95).
t
Hence, the experimentally measured values of the effective diffusivity
for activated carbons are consonant with a simple model assuming pore diffu-
sion as the rate-limiting step. The good agreement between the measured rate
data and form of the uptake curves based on the model (Figure 24) also
strengthens confidence in the validity of the model as a means of interpreting
our data. It appears unnecessary to consider surface diffusion as an impor-
tant phenomenon contributing to the internal transport of the broad spectrum
of organics measured by DOC. If surface diffusion were rate-controlling, the
experimental pore diffusion coefficients would be substantially in excess of
the molecular diffusivity.
The interpretation of the rate data in this work in terms of the pore
diffusion model for internal transport should not be construed as a rejection
of the surface diffusion model. Rather, it is intended simply to emphasize
that transport by pore diffusion appears to be significant, and that the pore
diffusion mechanism is sufficient to explain the observed rates, in view of
the limitations of the data. This is not meant to imply that the interpreta-
tion of the rate data in terms of pore diffusion is the only correct approach
to data analysis. Indeed, it is theoretically impossible to distinguish
90
-------
CO
EH
S3
M
M
P4
s
(~\
8
g
0
H
CO
t>
fe
H
O
O
w
H
^
H
CO
w
rH
pa
?<
MH
cy
Pi
CO
d
0
H
4-1
"rl 4-1
d fli
S g
t .
1 §
cd ;2
Co
CO &
C»
II
§
CO
^4
1 ^N
CU MH £}
0 0 W
P-i '-'CM
4-> d ^
d o ^
* *w -1-1
cd 3 g
ft'w .H
b £
4-1
d
cu
rQ
rJ
0
CO
3.
Vi
rH Hi
0 *
^p
60
d co
H CU
CO -H
3 o
MH CU
m &
H CO
o
CO
rH CU 0 CO ^4
4-1 3 O vH ^
CXrH J-l J3 O
X Cd MH 4-1 &
W >
pj
H
trj
CU
H
O
CO
cu
U
^
d
o
H
CO
3
MH
MH
H
«
CU 4-1
r4 ^
o cu
P-I 4-1
0 O O O
I 1 [ 1
1 1 1 1
O O O 0
rH rH rH rH
X X X X
CM VO rH si-
CO rH i-l rH
O CM O 2
^D *>^ CO TV
oT |3
60^
d & o '
cd ^ o
(H m
cu o A.
TJ in
rl
S 2
O
o
Q
X-N
sj-
z,
^
O
i
e 5
T3 O CU
cu o si
4-i ti coro
cd --.
*t? cd d I
&*» cd a
^3 M m
d cu T3i
3 4-1 -H O
C8 CU rH
CU S Pi
60 ^^ X
VI d
cd -H CU sj-
rH J-l P-.
co 3
>, CU 4-1 CM
H rH Cd
cu 3 Vi 11
> 0 CU
M rH 0
o o cu
P* e 4J
o o o
III
000
rH rH rH
XXX
m vo -*
CO rH
o
^4
**
H
TJ
o § |
O O "
m > /~T
l/"} ;vr
un
: CO
rH CU
Cd CU rH
O 60 3
H M CJ
o< cd cu
=H i
x^»
^"
2x
m
s~**
Tjg
o
o
s
d CO !-l
03 CU
H o cu ,d
CO CU 4-1 CO
3 3 -H
M-< V d Td
MH cd -H d
H m cd
o d d
H iH T3
MH *H
O CO 4-1 tt)
4J cd pi
cu d ^
rH CU CO
rQ -H d d
cd o o o
4-1 -H -H -H
MH 4-1 4-1
S MH 3 3
O CU H i-H
VJ O O -H
fi( O CO T3
O O O
1 1 1
000
rH i 1 i t
X XX
r-i co r<
CM in r-
r-4 t 1
H
CM CM ^.
CM o rf
rH rH "^
*^3 n
71 -H ^
8 S S-3
.y o M *o
S rH 0 rH
2 Cd CO Cd
§ X rH
PQ °
X-N
O
d
d- o
0 -H
CO
T3 "3
cu d MH
co cu MH
cd T3 iH
CU
" 4J CU
d 4-i n
O iH O 4J
H rJ P< d
CO O Cd
3 4-1 O
MH MH cd iH
MH O ,d MH
iH 4J . -H
TJ rH d
CU 60 60
CU T3 d . «H
CJ Q -H W
MH 3 -H
rJ CU CO
3 ^3 CO CO
CO 4-1 CS -rl
CM CO
1 1
O O
rH rH
X X
CO C^
in vo
co -a-
o
o
g
a a.
cu
d
3 I?"
S "° o
J? TK! ^H
cu M
o
o
^\ ^^N
i-l rH
VO VO
4-1
4J d
d cu
CU -H
H CJ
CJ -H
MH MH
MH CU
cu o
0 CJ
o
d
d o
O 'H
H CO
co 3
3 MH
l_j_ 1 III
MH iH
H X-N 'O x-\
T3 N N
4-1 t3 4-1
CU -H -H -H
Vl }J rH VJ
O F"4 O Pn
CM ^^ CO ^^
O CO
rH rH
1 1
O O
rH. rH
X X
m CM
i
rH rH
CM
O
rrj TJ
CU .-i CU (H
S^ ^-s
^4 ^^
s a
CU
tf 'j
-1 o
0 rH
x CO
p
x^\
CTs
^
d
o
cu
CO
cd
«\
d .d
o cu
H -H
CO CO
3 W
MH
MH MH
H O
rH
cu cu
M *n)
O 0
ON
rH
O
4-1
rH
rH
0
rH
T3
CU -
Activat
Carboi
CU
60 '
§
1-1
fi\
UJ
^t
*rH
*
CJ
o
91
-------
between the pore diffusion and surface diffusion for the case of linear equi-
librium (61).
The apparent success of the pore diffusion model in simulating the rate
of DOC uptake is surprising. Previous workers (61) studying the rate of
uptake of model substances such as phenol and paranitrophenol by activated
carbon have found internal transport to be more rapid than could be explained
by pore diffusion, and hence have invoked the surface diffusion concept. The
cause of the difference may lie in the different nature of the solutes: the
model substances (phenol and p-nitrophenol) are more strongly sorbed than is
the collective parameter. DOC. Hence, the equilibrium isotherm for phenol and
paranitrophenol is favorable (n » 1 in the Freundlich expression, Eq. 9) and
the driving force for surface diffusion is greater than for pore diffusion.
For DOC, the equilibrium isotherm is linear, and surface diffusion is less
likely to be important.
SUMMARY
Results have been reported for experiments dealing with pyrolysis and
activation of lignocellulosic materials, physical characterization of the
resulting chars and activated carbons, and evaluation of the performance of
activated carbon prepared from a lignocellulosic waste. Lignocellulosic
materials were found to behave in pyrolysis according to their initial com-
position. The chars formed have a substantial microporous volume and specific
surface, but no measurable pore volume in the size range essential for trans-
port of large molecules. Activation in a C02 atmosphere at 900°C serves to
enlarge the pores, and to create pore volume in a size range suitable as a
diffusion network for organic substances of the sort encountered in water and
wastewater. One such activated char prepared from prune pits demonstrated an
adsorption capacity and transport rate coefficient (effective diffusivity)
equal if not superior to the corresponding values for Filtrasorb 400, a com-
mercially produced, coal-based activated carbon widely used for water and
wastewater treatment. A simple model based on linear equilibrium and trans-
port by pore diffusion proved useful in interpreting the data for DOC uptake
by activated carbon.
92
-------
REFERENCES
1. Hassler, J. W. Purification with Activated Carbon. Chemical Publishing
Co., Inc., New York, New York, 1974. 390 pp.
2. Hey, M., H. Marsh, and F. R. Reinoso. The Adsorptive Properties of
Carbonised Olive Stones. Carbon, 11:633-638, 1973.
3. Marsh, H., M. Iley, J. Berger, and T. Siemieniewska. The Adsorptive
Properties of Activated Plum Stone Chars. Carbon, 13:103-109, 1975.
4. Mahajan, 0. P., and P. L. Walker, Jr. Porosity of Coal and Coal Products
In: Analytical Methods for Coal and Coal Products, Vol. I, C. Karr, Jr.,
ed. Academic Press, New York, New York, 1978. pp. 125-162.
5. Dubinin, M. M. Porous Structure and Adsorption Properties of Active
Carbons. In: Chemistry and Physics of Carbon, Vol. 2, P. L. Walker, Jr.,
ed. Marcel Dekker, Inc., New York, New York, 1966. pp. 51-120.
6. Brunauer S., P. H. Emmett, and E. Teller. Adsorption of Gases in Multi-
molecular Layers. J, Am. Chem. Soc., 60:309-319, 1938.
7. Sutherland, J. W. The Usefulness of Measurements of the Physical Adsorp-
tion of Gases in Characterizing Carbons. In: Porous Carbon Solids, R. L.
Bond, ed. Academic Press, New York, New York, 1967. pp. 1-64.
8. Lamond, T. G., J. E. Metcalfe, III, and P. L. Walker, Jr. 6A Molecular
Sieve Properties of Saran-Type Carbons. Carbon, 3:59-63, 1965.
9. Lamond, T. G., and H. Marsh. The Surface Properties of Carbon - II. The
Effect of Capillary Condensation at Low Relative Pressures upon the
Determination of Surface Area. Carbon, 1:281-292, 1964.
10. Atkins, J. A. Porosity and Surface Area of Carbon Black. Carbon, 3:299-
303, 1965.
11. Chiche, P., S. Durif, and S. Pregermain. Development of the Internal
Surface Area of Coals During Carbonization. Fuel, 44:5-28, 1964.
12. Cranston, R. W., and F. A. Inkley. The Determination of Pore Structures
from Nitrogen Adsorption Isotherms. Advances in Catalysis, Vol. 9.
Academic Press, New York, New York, 1957. pp. 143-155.
93
-------
13. Walker, P. L., Jr., L. G. Austin, and S. P. Nandi. Activated Diffusion
of Gases in Molecular-Sieve Materials. In: Chemistry and Physics of
Carbon, Vol. 2, P. L. Walker, Jr., ed. Marcel Dekker, Inc., New York,
New York, 1966. pp. 257-371.
I
14. Micromeritics Instrument Corporation. Instruction Manual for Model 2100
Orr Surface Area and Pore-Volume Analyzer. Norcross, Georgia, 1969. 55
pp.
15. El Tawil, M. M., and L. F. Brown. Changes in Pore Structure of a Devola-
tilized Coal Char upon Further Heating at Lower Temperature. Carbon,
14:132-133, 1976.
16. Marsh, H., and W. F. K. Wynne-Jones. The Surface Properties of Carbon -
I. The Effect of Activated Diffusion in the Determination of Surface
Area. Carbon, 1:269-279, 1964.
17. Scholten, J. J. F. Mercury Porosimetry and Allied Techniques. In:
Porous Carbon Solids, R. L. Bond, ed. Academic Press, New York, New
York, 1967. pp. 225-250.
18. Washburn, E. W. Note on a Method of Determining the Distribution of Pore
Sizes in a Porous Material. Proc. Nat. Acad. Sci. U.S., 7:115-116, 1921.
19. Dickinson, J. M., and J. W. Shore. Observations Concerning the Determi-
nation of Porosities in Graphites. Carbon, 6:937-941, 1968.
20. Lehninger, A. L. Biochemistry, 2nd ed. Worth Publishers, Inc., New
York, New York, 1975. 1104 pp.
21. Shafizadeh, 'F. Pyrolysis and Combustion of Cellulosic Materials.
Advances in Carbohydrate Chemistry, 23:419-474, 1968.
22. Tang, M. M., and R. Bacon. Carbonization of Cellulose Fibers - I. Low
Temperature Pyrolysis. Carbon, 2:211-220, 1964.
23. Bacon, R., and M. M. Tang. Carbonization of Cellulose Fibers - II. Phys-
ical Property Study. Carbon, 2:221-225, 1964.
24. Stamm, A. J. Thermal Degradation of Wood and Cellulose. Ind. Engr.
Chem., 48:413-417, 1956.
25. Tang, W. K., and W. K. Neill. Effect of Flame Retardants on Pyrolysis
and Combustion of a-Cellulose. J. Polym. Sci: Part C, 6:65-81, 1964.
26. Brunner, P. H. Examination of the Pyrolysis of Cellulose for the Produc-
tion of Activated Carbon from Wastes. Internat. Solid Wastes Assoc.
Bull., 23:2-10, 1977.
27. Broido, A. Thermogravimetric and Differential Thermal Analysis of Potas-
sium Bicarbonate Contaminated Cellulose. Pyrodynamics, 4:243-251, 1966.
94
-------
28.
29.
30.
31.
32.
33.
34.
35.
36.
37,
38.
39,
40.
41,
Hawkins, W. L. Polymer Stabilization. Wiley Interscience, New York, New
York, 1972.
Brunner, P. H., and P. V. Roberts. The Significance of Heating Rate on
Char Yield and Char Properties in the Pyrolysis of Cellulose. Carbon (in
press).
Masters, K. J., and B. McEnaney. The Structural Evolution of a Micro-
porous Carbon with Heat Treatment. Paper presented at 14th Bienniel
Conference on Carbon sponsored by the American Carbon Society, The
Pennsylvania State University, 1979. (Proceedings in press.)
Freeman, E. M., and H. Marsh. Selective Structural Gasification from
Paracrystalline Carbon - A Critical Assessment and Gas-Phase Adsorption
Investigation. Carbon, 8:19-30, 1970.
Sarkanen, K. V., and C. H. Ludwig. Lignins. Wiley-Interscience, .New
York, New York, 1971. 916 pp.
Shafizadeh, F., and G. P. McGinnis. Chemical Composition and Thermal
Analysis of Cottonwood. Carbohyd. Res., 16:273-277, 1971.
Shafizadeh, F., and P. P. S. Chin. Thermal Deterioration of Wood. In:
Wood Technology: Chemical Aspects, I. S. Goldstein, ed., A.C.S. Symposium
Series 43. American Chemical Society, Washington, B.C., 1977.
Philpot, C. W. Influence of Mineral Content on the Pyrolysis of Plant
Materials. Forest Science, 16:461-471, 1970.
George, C. W., and R. A. Susott. Effects of Ammonium Phosphate and
Sulfate on the Pyrolysis and Combustion of Cellulose. USDA Forest Ser-
vice- Research Paper INT-90. Intermountain Forest and Range Experiment
Station, Ogden, Utah, 1971.
i , *
Rothermel, R. C. Forest Fires and the Chemistry of Forest Fuels. In:
Thermal Uses and Properties of Carbohydrates and Lignins. F. Shafizadeh,
K. V. Sarkanen, and D. A. TiUman, eds. Academic Press, New York, New
York, 1976. pp. 245-258.
Susott, C. W., W. F. DeGroot, and F. Shafizadeh.
Fuels. J. Fire Flammability, 6:311-315, 1975.
Heat Content of Natural
Pass, S. M., G. V. Desai, L. F. Brown, and R. E. West. The Consequences
of Different Temperatures on Pore Structure Development in Carbon.
Carbon, 12:619-631, 1974.
Lamond, T. G., and H. Marsh. The Surface Properties of Carbons - III.
The Process of Activation of Carbons. Carbon, 1:293-307, 1964.
Dubinin, M. M., G. M. Plavnik, and E. D. Zaverina. Integrated Study of
the Porous Structure of Activated Carbons from Carbonized Sucrose.
Carbon, 2:261-268, 1964.
95
-------
42. Yousef, A. M. Fine Structure of Lignitous Coals. Carbon, 13:1-6, 1975.
43. Miura, S., P. L. Silveston, and K. Hashimoto. Analysis of Pore Develop-
ment Processes During Gasification of a Carbon Char. Carbon, 13:391-400,
1975.
44. Mahajan, 0. P., R. Yarzab, and P. L. Walker, Jr. Unification of Coal-
Char Gasification Reaction Mechanisms. Fuel, 57:643-646, 1978.
45. Juntgen , H. Manufacture and Properties of Activated Carbon. In: Trans-
lation of Reports on Special Problems of Water Technology, H. Sontheimer,
ed. EPA-600/9-76-030, U.S. Environmental Protection Agency, Cincinnati,
Ohio, 1976.
46. Dutta, S., Y. Wen, and R. J. Belt. Reactivity of Coal and Char - 1. In
Carbon Dioxide Atmosphere. Ind. Engr. Chem., Proc. Des. Div., 16(1):20-
30, 1977.
47. McEnaney, B., and N. Dovaston. The Development of Porosity in Heat-
Treated Polymer Carbons upon Activation by Carbon Dioxide. Carbon,
13:515-519, 1975.
48. Tomkow, K., T. Siemieniewska, F. Czechowski, and A. Jankowiska. Forma-
tion of Porous Structures in Activated Brown-Coal Chars Using 02, C02 and
H20 as .Activating Agents. Fuel, 56:121-124, 1977.
49. Turkdogan, E. T., and J. V. Vintners. Effect of Carbon Monoxide on the
Rate of Oxidation of Charcoal, Graphite and Coke in Carbon Dioxide.
Carbon, 8:39-53, 1970.
50. Rand, B., and H. Marsh. The Process of Activation of Carbons by Gasifi-
cation with C02 - III. Uniformity of Gasification. Carbon, 9:79-85,
1971.
51. Walker, P. L., Jr., M. Shelef, and R. A. Anderson. Catalysis of Carbon
Gasification. In: Chemistry and Physics of Carbon, Vol. 4, P. L. Walker,
Jr., ed. Marcel Dekker, Inc., New York, New York, 1968.
52. Walker, P. L., Jr., L. Pentz, D. L. Biederman, and F. J. Vastoka. The
Influence of "Inert" Diluent Gases on the Rate of Carbon Gasification.
Carbon, 15:165-168, 1977.
53. Marsh, H., and B. Rand. The Process of Activation of Carbons by Gasifi-
cation with C02 - II. The Role of Catalytic Impurities. Carbon, 9:63-77,
1971.
54. Hennig, G. R. Catalytic Oxidation of Graphite. J. Inorg. Nucl. Chem.,
24:1129-1137, 1962.
55. Langmuir, I. The Adsorption of Gases on Plane Surfaces of Glass, Mica
and Platinum. J. Am. Chem. Soc., 40:1361, 1918.
96
-------
56. Hiemenz, P. C. Principles of Colloid and Surface Chemistry.
Dekker, Inc., New York, New York, 1977. 516 pp-
Marcel
57. Jain, J. S., and V. L. Snoeyink. Adsorption from Bisolute Systems on
Active Carbon. J. Water Poll. Control Fed., 45:2463-2479, 1973.
58. Weber, W. J., and J. C. Morris. Equilibria and Capacities for Adsorption
on Carbon. ASCE, J. San. Engr,, Div., 90:79-107, 1964.
59. Weber, W. J. Physicochemical Processes for Water Quality Control.
Wiley-Interscience, New York, New York, 1972. 640 pp.
60. Mattson, J. S., and H.B. Mark. Activated Carbon Surface Chemistry and
Adsorption from Solution. Marcel Dekker, Inc., New York, New York,
1971. 237 pp.
61. Fritz, W., W. Merk, and E. U. Schlunder . Competitive Adsorption of Two
Dissolved Organics onto Activated Carbon. Part II: Adsorption Kinetics
in Batch Reactors. In: Proceedings on Practical Applications of Adsorp-
tion Techniques in Drinking Water, NATO Committee on Challenges of Modern
Society and U.S. Environmental Protection Agency, Reston, Virginia, Apr.
, 30-May 2, 1979. (Proceedings in press.)
e
62. Freundlich, H. Colloid and Capillary Chemistry. Methuen and Co., Ltd.,
London, 1926
63. Sontheimer, H. Basic Principles of Adsorption Process Techniques. In:
Translation of Reports on Special Problems of Water Technology, Volume 9
- Adsorption. EPA-600/9-76-030, U.S. Environmental Protection Agency,
Cincinnati, Ohio, 1976. pp. 29-66.
64. Dobbs, R. A., R. J. Middendorf, and J. M. Cohen. Carbon Adsorption Iso-
therms for Toxic Organics. Municipal Environmental Research Laboratory,
Office of Research and Development, EPA-600/880-023, U.S. Environmental
Protection Agency, Cincinnati, Ohio, 1980. 321 pp.
65. Radke, C. J., and J. M. Prausnitz. Thermodynamics of Multi-Solute
Adsorption from Dilute Liquid Solutions. Journal AIChE, 18(4):761-768,
1972.
66. Fritz, W., and E. U. Schlunder. Competitive Adsorption of Two Dissolved
Organics onto Activated Carbon, Part I: Adsorption Equilibria. In:
Proceedings on Practical Applications of Adsorption Techniques in Drink-
ing Water, sponsored by NATO Committee on Challenges of Modern Society
and U.S. Environmental Protection Agency, Reston, Virginia, Apr. 30-May
2, 1979. (Proceedings in press.)
67. DiGiano, F. A., G. Baldauf, B. Frick, and H. Sontheimer. A Simplified
Competitive Equilibrium Adsorption Model. Chem. Eng. Sci., 33(12):1667-
1673, 1978.
97
-------
68. Crank, J. The Mathematics of Diffusion. Oxford University Press, Amen
House, London, 1957. 347 pp.
69. Roberts, P. V. The Adsorption of Normal Paraffins from Binary Liquid
Solutions by Molecular Sieve SA Adsorbent. Ph.D. Thesis, Cornell Univer-
sity, Ithaca, New York, 1966. 230 pp.
70. Crittenden, J. C., and W. J. Weber. Predictive Model for Design of
Fixed-Bed Adsorbers: Parameter Estimation and Model Development. ASCE,
Jour, of Env. Eng. Div., 104(EE2):185-197, 1978.
71. Crittenden, J. C., and W. J. Weber. Model for Design of Multicomponent
Adsorption Systems. ASCE, Jour, of Env. Eng. Div., 104(EE6):1175-1195,
1978.
72. Weber, W. J., and J. C. Morris. Kinetics of Adsorption on Carbon from
Solution. ASCE, San. Eng. Div., 89(31):31-59, 1963.
73. DeWalle, F. B., and E. S. K. Chian. Removal of Organic Matter by Acti-
vated Carbon Columns. ASCE, Jour. Env. Eng. Div., 100(EE5):1089-1105,
1974.
74. Parkin, G. F. Sources and Characteristics of Soluble Organic Nitrogen in
Activated-Sludge Effluents. Ph.D. Thesis, Stanford University, Stanford,
California, 1977. 245 pp.
75. Keller, J. V. Activated Carbon Adsorption of Soluble Organic Nitrogen
Compounds in Municipal Secondary Effluent. Engineer's Thesis, Stanford
University, Stanford, California, 1976. 55 pp.
76. McCarty, P. L., D. G. Argo, and M. Reinhard. Reliability of Advanced
Wastewater Treatment. In: Proceedings of Water ReuseFrom Research to
Application. Amer. Water Works Assoc. Research Foundation, Denver,
Colorado, 1979. pp. 1249-1275.
77. Sontheimer, H., E. Heilker, M. P. Jekel, H. Nolte, and F. H. Vollmer.
The Mulheim Process. J. Am. Water Works Assoc., 70:393-396, 1978.
78. Frick, B. Prediction of Multicomponent Adsorption Behavior in Activated
Carbon AdsorbersEquilibrium Aspects. In: Proceedings on Practical
Applications of Adsorption Techniques in Drinking Water, sponsored by
NATO Committee on Challenges of Modern Society and U.S. Environmental
Protection Agency, Reston, Virginia, Apr. 30-May 2, 1979. (Proceedings
in press)
79. Hsieh, J. S., R. M. Turian, and C. Tien. Batch Adsorption Kinetics in
Sewage-Activated Carbon. J. ASCE, San. Eng. Div., 97(SA4):505-521, 1971.
80. Levy, S. J, San Diego Demonstrates Pyrolysis of Solid Waste. EPA-
SW80d.2, U.S. Environmental Protection Agency, 1975.
98
-------
81. Klumb, D. L., and P. R. Brendel. Union Electric's Solid Waste Utiliza-
tion System. In: Energy and Resource Recovery from Industrial and Munic-
ipal Solid Wastes, G. F. Kroneberger, ed. AIChE Symposium Series #162,
Vol. 73. American Institute of Chemical Engineers, New York, New York,
1977. pp. 160-167.
82. Moore, W. E., and D. B. Johnson. Procedures for the Chemical Analysis of
Wood and Wood Products. Forest Products Laboratory, Forest Service, U.S.
Department of Agriculture, 1967.
83. The Chemical Rubber Company. Handbook of Chemistry and Physics, 48th
Edition, R. C. Weast, ed. Cleveland, Ohio, 1967-1968.
84. Dohrmann Division of Envirotech Corporation. Dohrmann DC-50/52 Series
Total Organic Carbon Analyzers Equipment Manual. Santa Clara, Cali-
fornia, 1977.
85. Pacheco, L., D. S. Yan, and H. Marsh. Activated Carbon from Sport Coffee
Grounds. Paper presented at 14th Biennial Conference on Carbon, spon-
sored by the American Carbon Society and The Pennsylvania State Univer-
sity, 1979. (Proceedings in press.)
86. Boucher, F. B., E. W. Knell, G. T. Preston, and G. M. Mallan. Pyrolysis
of Industrial Wastes for Oil and Activated Carbon Recovery. EPA-600/2-
77-091, U.S. Environmental Protection Agency, Cincinnati, Ohio, 1977.
170 pp.
87. Dandy, A. J. Production and Characterization of Activated Carbons from
Agricultural Waste Products and Wood Charcoal. New Zealand J. of
Science, 20:291-294, 1977.
88. General Electric Company. Solid Waste Management Technology Assessment.
Van Nostrand Reinhold Company, New York, New York, 1975.
89. Mackay, H. A. The Influence of Polymer Structure on the Conversion of
Synthetic Resins to Carbon Coke. Carbon, 8:517-526, 1970.
90. Sawyer, C. N., and P. L. McCarty. Chemistry for Sanitary, Engineers.
McGraw-Hill Book Company, New York, New York, 1967. 518 pp.
91. Gan, H., S. P. Nandi, and P. L. Walker, Jr. Nature of the Porosity in
American Coals. Fuel, 51:272-277, 1972.
92. Roberts, P. V., J. 0. Leckie, and P. H. Brunner. Pyrolysis for the
Production of Activated Carbon from Cellulosic Solid Wastes. In: Solid
Wastes and Residues, Conversion by Advanced Thermal Processes, J. L.
Jones and S. B. Radding, eds. ACS Symposium Series 76. American Chemi-
cal Society, Washington, D.C., 1978. pp. 392-410.
93. Dixon, W. J., and F. J. Massey. Introduction to Statistical Analysis.
McGraw-Hill Book Co., Inc., New York, New York, 1957. 488 pp.
99
-------
94. Reid, R. C., and T. K. Sherwood. The Properties of Gases and Liquids.
McGraw-Hill Book Co., Inc., San Francisco, California, 1966. 646 pp.
95. Satterfield, C. N., and T. K. Sherwood. The Role of Diffusion in Cataly-
sis. Addison-Wesley Company, Reading, Massachusetts, 1963. pp. 20-23.
100
-------
APPENDIX A
USE OF LINEAR REGRESSIONS FOR DATA PLOTTING
In this appendix the justification is presented for reporting ash-free
char yield versus Tp and carbon yield versus Tp as linear regressions of the
experimental data.
Figures A-l and A-2 are plots of average ash-free yield versus T-r, for the
lignocellulosics discussed in the body of this report. Data points are con-
nected by smooth lines, resulting in curvilinear plots. Table A-l is a sum-
mary of linear regressions of the data plotted in Figures A-l and A-2. Figure
A-3 is a plot for four of the least linear (lowest r ) ash-free yield versus
Tp linear regression relationships along with the data points and their 90%
confidence intervals (standard deviations having been estimated as footnoted
in Table 3 of the report). Data at 600°C were not duplicated, and hence no
confidence intervals are shown though it is reasonable to expect confidence
intervals of roughly the same magnitude as the other data for the given mate-
rial. From Figure A-3 it is clear that statistically the data do not differ
significantly from the linear regression. Thus only the linear regressions
are used for comparative purposes in the report.
The case is similar for carbon yield versus T^. Table A-2 lists the
results of linear regression of the data. Figure A-4 displays the linear
regressions, data points and confidence intervals for two of the materials.
Again the data do not differ in a statistically significant way from the
regression, and therefore only the regressions are used in the report for
comparison.
101
-------
o
ti50]-
CC40
UJ
UJ
0:2
U-
I
0
I 0
-o
VPECKY CEDAR (^
o REDWOOD
DCORN STOVER <£
-------
o
Uj30
F
tr
< 20
O
(T
U.
i
*v
350C/min
o REDWOOD
O CORN STOVER
D COMPUTER PAPER
0 ' 500 700 900
FINAL TEMPERATURE (°C)
Figure A-2. Ash-free char yield versus final temperature:
rates.,
high heating
103
-------
90%CONFIDENCE INTERVALS
50
40
30
cr
x
o
LU
UJ
01 30
VPECKY CEDAR
O REDWOOD
0
500
700
I
CO
20
10
-Ar
QHSr
15
15
900
D CORN STOVER 15
OCOMPUTER PAPER 15
Figure A-3.
0 ' 500 700 900
FINAL TEMPERATURE (°C)
Linearity of the,plots of ash-free yield versus final temperature
for pyrolysis at 15°C/min
104
-------
90% CONFIDENCE INTERVALS
LU
>-
O
CD
60
50
40
oo REDWOOD
O--O COMPUTER PAPER
500
700
900
FINAL TEMPERATURE (°C)
Figure A-4. Linearity of the plots of ash-free yield versus final temperature
for pyrolysis at l°C/min
105
-------
TABLE A-l. SUMMARY OF LINEAR REGRESSIONS OF ASH-FREE CHAR YIELD
VERSUS FINAL PYROLYSIS TEMPERATURE FOR SELECTED LIGNOCELLULOSICS
Material ,
Pecky Cedar
Redwood
Corn Stover
Computer Paper
aRegression Equation:
expressed in °C.
TABLE A-2.
°C/min
1
15
1
15
Max
1
15
Max
1
15
Max
al
-0.03
-0.02
-0.02
-0.02
-0.02
-0.01
-0.01
-0.01
-0.02
-0.01
-0.01
Regression Constants3
ao
62.50
58.89
46.32
40.09
36.50
35.80
30.30
29.15
37.35
25.59
24.00
r2 -
0.96
0.89
0.95
0.76
0.90
0.98
0.77
0.90
1.00
0.85
0.97
Ash-Free Char Yield (%) = axTp + aQ , where Tp is
SUMMARY
VERSUS FINAL PYROLYSIS
Material 4> ,
Pecky Cedar
Redwood
Corn Stover
Computer Paper
Degression equation:
expressed in °C.
bOnly two points (500
°C/min
1
15
1
15
1
15
1
15
Carbon
, 700°C)
OF LINEAR
REGRESSIONS OF CARBON YIELD
TEMPERATURE FOR SELECTED LIGNOCELLULOSICS
al
-0.021
-0.009
-0.015
-0.010
-0.022
-0.004
-0.023
-0.011
Yield (%)
available
o
Regression Constants
ao
85.5
74.8
68.3
58.4
67.1
51.0
69.7
50.1
= a-^Tp + aQ , where Tp is
for "regression."
r2
1.00b
0.69
0.84
0.70
0.99
0.76
0.91
0.92
106
-------
APPENDIX B
MERCURY POROSIMETRY MEASUREMENTS
The mercury porosimetry measurements, as determined by American
Instrument Company, are shown in the following Figures B-l through B-7 for
42M, 30M, 15M, CHAR, F400, F100, and AN-A. These have not been corrected for
the effect of external porosity.
107
-------
«
1
j
o
k»
s
s g
f. K
UJ H
a o
>
H
w « 1-
2 ^i ^
GC -T t
0 =
3
UJ U
O
§
1
3
o
^
g
|
o>
s* a.
'? *
£
K
° £
z i
0; 0
1U i
0 -
o:
O K
UJ
K
UJ
2
«f
O
UJ
c cp o
3 1 a.
i 7 t
i ^ *
l-l "I
J H
5! 5! uj >
3 :
< «
5 < =
t ^ a
v> v> a u
Is
o .
Is
0.002-
,003-
.004-
.005
,OC6
.037 -
,OC»
.009
.010
0.02
0.03-
0.04-
o.os
0,06
O.OS
olio
0.2
0.3
0,4 -
05
0,6-
07
0
09
1 0
2
3
4
5
6
7
1
9
10
20
30
40
50
60
T;?.,-T.|-., v.-->r.r?.:-F' ''!:-
t r r-, : : i j-j-i-f r r :.-! !"
tLLini.lirpuL.cni!
^Win!h-iH!v?:\n
.r r ,.,-. .j.i, .-iyj--t ;-t-fT'..]': '\' '
TTTTTTfrfrrcnTnrr^''
r 1 :<< \ ' ', < t V r-j-1-r-t1)- '
I !"i T'i 4 ' { ; fj"| j'f.il :"! J : : . ' '
":": . . i' : .-;,; j- : : ' : : :::;.'
i i i ; ; i"i- \ ; f i ; rrH'H '.- -* -j
i i M!!..; ; ; | ;.L t.i 1.;. <_j U.i : . i
-r-i;. { ! : H-i'hr: H't i rr i rr.i r !VT H i M i [I r|.| fTi
Lii.^.t.L:_L44iU-: j.:.,:-:-i.i4
rH^HtlT :';:T|:r:::iu
^TJ(LiTh-n.rn'riTr:!.l:
,,;-. ;{:,., H-i-r i i i . :-: r ! -
f.{ j.s| j...|.j.!L.i.t |.j ].| .: |.| 1
Ti-pHTrlTnTTrfF-}
-i_L .. -: 1; ' > ' i -r ' I i i" <; !
l T . . , ill 1 f ; :,, i ,! i. i ,- : - !
.: ; - . i . i
* : -: i
iitt!
TT . "'
,i..i-.i .. . _ , - .. i -r- - , ' i '-[- T '
[.rl! i'il.iT
t: | i j f : H
: : T'-" j T ; :";'
_;::[::.::
ill
- r~ r ;' 'i y ^ * ^'T j'
"T~r r i": ; r ~T *
IJjTjT ;.[}"
: hi : :
.,.*. i j ,- . , = r
rariiii.;-].
i : i i i i 1 : i I
Sit
(" 1" : . '; i $ !
~ T*l ~!"T ?~" *J 1 f
tm-kMir
~i I" ; ' " " r "''' T t" T r ; i ~ T " * i -. "
"T~; ?' ' " ', r ;i t ? r ~ i i r i i'i j "
=h^"H *r-'fi7~H'T rH ! ~ r;*r1
_1 i-' i»t_; : ;.,:.1. 4. j- :. j -j ; r : i
*;;;:£-»::::!:::;;
jiiriLUJ-
* ;^ ;r ; * "t
; v_; t : : r:
"t '' i j r"f~^"^~r
Tifuiub
^.^..1.1 ... J_i_
= ) 1 -I .; .4 -f i j. j-
"" "if.: ' I.J.
_^ .. 1 J ! ; . i i
4.4.1! h.l U.
* " I ' * ; ;_
~. ... t ....
;:::{::.:;
4'L^-rrf ' T r
::::ji;:rj
'. '. l ^ ', , ' . '. '. t
: : 1 : .' ;J
. t j * J.j r [ ' . j
It:;.! l.JT! ! !
.J. .^_. . i.l- i : S i t-
-; - i T -j-i : j- j t
.1 _ ! S. '>' ' .. I !
\ i 1 ,,i J -.-( 1
rr ; 1 r. t M r
~ T".1 .* r~- "* -'* TV
: : : : j : : : i
,*j . i : | *';"?; "T~* "it"*" r :
r - . r ! t r '^ . ; j
* r T "_' . ;""":'! ! f : i ! i 1 '.
j t f ? : ; "--' s-r j1 r- -": *:* t _ . : \
^: . ; . : :"i"r ;'':* -1: " n
-*:''-,'. i '!:!:-. : I
^-H-ri j'rr!"l"t"i I'fl i 'r^ H
i i - ^ : ! ; i ;! M i ' fs ' :
l-M-s i-ji-v; ! 1 i r;f r; i n ":
LLLLlil.;i.;'Ll.nnB-!.i.= -; 1-
r ^ : i::in; ;ni j, | j : : : t r ; j
I::::;:::*.;:: :.: i r"r t
iSlll^j-H-i-RiSi-li:
; :
T) i-:.
, VI « : ; ;!l. ' ; : , i^i 1 ^ ;_T .i ,j-j .,'-.: i
t|.i.I!JI TU-U"; r H-H-t ! f T L :
*-=rrH ^l.U r i "nirtriji?' ;
- rX - ''h ^""^ t :- **> !::T:r [-i-f !*-' ;'
~f ..^ I,,' ; j,| 1' '. 1 ; ! : J ' ;' j t "' i ; i . 1
r"Mv"T lt~-~1~rn y1~1~-": ": r| rj r
f- : c *i"r~:i*-r :'' i ; r~r-; : -r1 r : -f : : :~?"
. ; :$.":'::.: ; : : : ' : : ; t ; "
;.4_ ' "" * * * ^ '. ; .U -U " .^_L....^ 1
1z r t t*- *»"*
,_, i/.'4-i U.' r..;::..L:-;.i-
. . , .- .
.;'.;."..
; j ,( j. Jt I;_4-, f. j ; ;, i | .-. I |-t- i -; [ | i .,
;. .. j i. ,:^ . . *.. ; .'. .. , i- j ;. -T -. 4 ;-}-; ; - >
;,;,i_li £'- ;:.\|>rrr- 1 T : r ! ? : 4-T'T-r f-! :' " ""
"" !~t , ! t; r- j- , ! f j"f ' J" f f T T'
! : IIL' r I; i ! :,^ j ; ivj-^' tt'-i-
V : :
--r i .
~r I i": ! "; * - 1 TIT! ! |T i : .: "
.fl-j-.^-JM
I . ! 1
-j"
I '. : :
Mr
i ;4±
S-:-!'l
|i.l.i-
-:: ;rf.
;il
.rrrr
i ; j-h
:' ,r~f
: j T
; , T : «
!ii
' !.-
-:-: "^i"
t ! 54-
; i.t-1.
m
.l:tf
if
' - '"
...i ..;«.
ij-'rJJ:
i . «
;. 1 .!.: S '-
;,;«.'
I! 1:3^
r r~rr Q *
' * -4" g ~
yjf"!
[ i * ^-j-
--t _**-!
i ' ' '''}
['rJ-h
; 4- 1 1
* ' *' r~i
' ' - ! "I
i i *:'?
i : t ;.'
r-r-r-rrr-r r {>. j^^rrTTJT ~- v'H-
::::.;:..:.... l-pTTT-l L
r in 'ri i-! !_', ! ; .i.r-'H "JT
' : - * : '.["! , . ; ,
:h '..'
i ;
"'*:'; : .'' . ..i. i.. '-. _J..; . '
; " 1 ! i ": . ,' ! I I I ]_ . '. '
. ' I T": : I"1 ""'" ' ; ' ~~ . : ~ ; i 7 . -
\ i'T'i: i :" ' :; i-^Vfrn ; ; :'~
' ' l' 1 t . }
~ "T *t ' : T r i i , j ,, ', "I/, I' * t ' J
: ;- t 'r-: "1 r-; = . :",;- '_'} i- j"; f-i-
jp; :j4|:im 11;] {niJjiff mi m ifjM
rrT" r*"rrrrT:' : : : r"' . TTT
l.rr ::::;:!. ..:.....
jj1-; 1 i ""~ '' \ ';'.','. i. '.'.'.'. \
o ' " *i vo -.1 <
r - / '* i !"T
ENET ATION. cc a t ^ _- __ »
i °. S '".
! ! . '
. ;..:
"H'H
^Ili
M r* .-1 r~t r4 i-l O O O
i. i f-
^ » i t-
' -i
i i.
1 '.* : i
it::'
'' ",! ,"
- i :
T !.
i ? '-. r
iLiii
fifti
I I'lT
rlj-p
li
00.000 fN
.600 *
.000
.000 M
,000 O
.000 ""
olooo M
0,000 »
0.000 *
5.000
0)
o.ooo o
.000 g^
.000 *^
??? CO
,000 i^
,060 M
,000 Q;
,000
(L)
.000 3
,000 0
oo O
00 -H
700 J-)
oo (0
500 £
400 £
c
10A <"
300 j^
200 g
^
H
.U
CO
"S3 s
90 3
to g
70 Jj
60
50
40 >,
J-l
30 -U
CU
10 -H
03
O
O
10 <^
7
t ^^
7 t-l
4 3
S Jj
4 Q)
^
I
2 -f
0)
V-i
f.1 S,
ot> .H
Pr.
108
-------
ec
ul
o
e"-.
a.
<
i-
o
UJ
a.
<
o
. 'Z.
a:
O
§
13
0.004
0.005
0.006
0.007
0.008
0004
0.010
HTf-' ri#tf tlfrFTiTntH Ifi'H ^
-
; rr-r+f-rf-H-j-rrf !-f-i--t-r-< t M ; M rtlr: rt^S :'i i H-h-rhrf r r
i~t~5"?"t~?~rr1~rrT"?Ti"ri'M~rf"7TrrTl~r1~j~ttT"^H' :;-;-*-:. s j ;->-!+-;-j-
-H-H-M-H-H != -H-H-M-H-^- 30.000
20.000
15,000
10.000
9.000
8.000
7.000
6.000
5.000
4.000
2.0OO
l.OCO
too
800
700
600
500
400
H^HH-;u.;::.rfLH..;H'~~i:
50-f-F
60 -I'".
70 44-
J8J5-
100-f-i
"-;-r-ri ; i- j-r-;-r-: 4- f i-; . . j.:.-;..;_.; f.;_r-;_!
.!irj_L! ' I.-'! ;.'jj.t r J = i j i !_i:i_| { 't ;!
t-f-T" ~r *' r ^-r-! "LT"^7~i'
NETRATION. cc/s f_'. -L i ;
i J : : I j t-Kt;
li'rf T
%: ve
cJ cf
CJ
cf
100.000
90.0OO
80.000
70.0CO
60.000
50,000
40.000
100
90
80
70
60
50
40
30
CO
Vl
o
OJ
4J
0)
T3
Ol
O
CO
CO
h
s
1
a
o
r-l
4J
(0
M
JJ
a
OJ
a)
t-H
g
4-1
I
CO
o
n.
o
ft
o
cf
ts
0)
j-i
S,
H
109
-------
II!
Ill
O
tc
o
o
K
UJ
Q
a:
O
r-l
C5
a.
<
UJ
(-
i
<0
~l u |. - T|[
j.ooj -I.' r ' . ' H
ooo
0,005 -
0,006
O.OC7
0003
100J
0.010
0.03 -
t-i"l.; r-j j m !{: !! 1
'o6 -t-ili rntrl il :; i r; nl't? \\ \ I :|. ! 111 j ?T;; i T |"
,.o7--,-:-H-:trrf j ! 1 : 1 rHi Hr: rj--?: r m;! [ rf [
1.08 -i ::«:::: :_{_: : : :}:::: {-; : : : I : .t_L_; .J_Li,:..iJ-:-J_: -i-
o.s -
06
07
0«
09
I 0
Ti-Tri'r.1 r.f
h ti-l i , i~!
rn-s.riTrr
'ffrffifff
TT1! r-j-rr H''JTi~r r f-i-r'S*
* : 1-1 -; t t i j ! ! j i i ;-;3-
n'|trrirrn-hr: rrn^rfttl-
i i_j__ ui _ .
Q^ TTf; j j j i; i: iS4jJJ_U i [_Q jj ! i! JJ j I
00.000
0,000
0.000
0,000
0.000
o.ooo
0.000
0.000
0.000
5.000
0.000
9.000
8.000
7.000
6.000
5.000
4.000
1.000
2,000
1.000
900
100
700
600
500
400
300
200
o
M-1
$-1
CO
4J
I
rl
J-l
o
ft
CO
3
CO
P
81
0
O
4-J
Ctf
0)
(3
-------
-jf
e-
3S
X
o
o
u.
o
* 2
a.
X
UJ
UJ
_j
a.
tu
<
§4
33
"
rrniti/riitiFriii ft ! r rrTt iifjtfg?
3,
111
-------
I
ui ja
0 &
s
o
ce.
a:
o
a
u,
UI
-I
O-
<
Ul
1
<
i
u
£
<
Q
<
>
3
a
mijihimimm^mU-U'imi
^^f nit rnrjuBTn.nU; ^fwff
:T:FrfTF!TT[|rrT}iTHiTiTl|lf]:r
-f i-m-Hi+rt-H+rr. n ! fH'f th'!
h rhtrj"- H-pvn-tr;; i.rrMTrrn T
TOijMii[Hi:Hppj
0.4-
0.5 -
0,6-
07
08
09
1.0
--..,,« 100.000
-.' - ivo.ooo
| 1,0.000
. : - j xo.ooo
.. i . 1 : j 60,ooo
: I :i ; j-l50.000
0.000
0,000
0.000
5,000
0.000
9.000
.OOP
'.000
6.000
5.000
4,000
3.000
2.000
^Ittltiil-tlllMitilt
^iitTnfHm
s- : -i-s-:.-.-i i = '"1 t-rr*' - t-i-r
liiHittrrr; t l^rrillt z'"
^:^^
7- -.
I ---
94-!
i '" it: T ! i ; ' ' '
. , -. |-i-S-r5t~-!"-hf rn-r
.1 i.|4-j J.M.I ;.; j : i.i-:-i-i-
4--\! ;! !: rl- H W 1 i I i-ii'M:!TrH't
: : : : : : : : ::::{::
: :^; r ;T{-_
TFiTiHF
.000
900
ooo
/oo
600
500
400
300
100
90
80
70
60
50
40
39
o
o
Vl
o
a)
4J
0)
H
13
0)
h
o
01
3
CO
I
a
o
H
4J
01
13
CD
p.
t-H
1
o
0}
o
O
cx
PQ
60
H
112
-------
POROSiTY DETERMKUTICN
cSM&A SAMO|I. Filtrmsorb 100 34-75M
O.U13
SAUDI P wpirtur -. .
ll-2«-7S XS-3V7*
DATF OPHPP M«
EQUIVALENT PORE DIAMETER (MICRONS) . 17S/PSI (CONTACT ANGLE = 130")
O.O02
0.00 J
0.004 -
O.O05 -
0.006
0.007
0.003
0.001
0.010
0.02 -
0.03 -
0.04 -
0.05 -
0.06
0.07 -
o.oa
0.09 -
0.10
0.2
0.3 -
0.4 -
0.5 -
0.6 -
07
0.8
09 -
1.0
2 -
3 -
4 -
5 -
6 -
7-
i -
10 -
20-
30-
40 "I
7°j
loo-;
J
!'!l-il:-:?;:' :;;;i
-I r T -:::::
ri -.T-.-r-. :: : t
.;\':; j;; ;
1 H 1 !-! P i : ^ l-i f
-iinmr -i ;l;tr
.iNm-i
: [" F' , ; ; i : i ''
i , : ! I
: ' ) :'' '-
TV:'Y n r.= n ; ni"'T <
i'*' ::::::: i : : : : i 1 : : :
:* ; )"! il n~! '
L-^LLU
(- i -. . ; . ? '- : r-:-; ; . . . ;;;-
! 1 ! i -i i L
[ ! i i j [ ,
1 ::::::::
[::::!::::
~:rTi T4"r"i Tl
S-ili-N IH-i
j !::;:: = : I::
liitilrn Hit-
-f .-.-,.,.. f >-r r^'-r
l i-l M ! . : I >-t
' ! ; fl - 1 !'H i 1 I-H. r:-! h-
: t ; ; | !
! ! H N
{.:'!'!! .''i'TTI
i ! ! 1 ! ; H -t!
r! r i H ! ' rrr
liiii in ?!! = !! nrrt
....... 4 .(....},... ,t_t-
i : :f If ff ffi'i ' j'j-fh HI" i ' j rh:
! : ;Tr- r -ji-l-t n-jrti :1 r rr= n r
...:.. . - J l :..L ..-; ; f ; i-r-1-; r i - : i ',- -r , r~i-
- 1 i ? mi; ! i : if! ;! Pi.i i ! -i=-nn! 1 1 M !!}!
n.nyrrn:r-i : ITT nrtn rrr ' r?lTTrn rvi"; rrr
Ti'rtttttrrtitrMti
_1_; ^_»_i t .1..;..: i.
T: :::::::.:
r-H-rr-r-r : ff «-H-r?T-:=~ H-rrrf
:-:::t : : : i : : :.:ti : : : : 1 : ;: ;
L{FI]tiTi:rT:T!T10!Til;:T TU
i i f--:- f- j-1- =-: : : 1-' i- -*'- '* r J
f ; : = : : j ; ' : : ;.;: '.
- 1 i I I i- J -4-
' i. ; I.1..1. ;._i_l
r : ; : : I : : : :
j. [ }_[ HlfJ;
,. ; -,.,-!. I-T j-|-^
-i ::::::
- - : -r i :- : - -
: : l : :
. .: - |. ^ .;. .
' '
i i ! : i
« i i r * t i ' ^_
+ (-* 'ft
; ! i
lilt! ;
4 I ;
-kt-i-t
.-X-1-- -:
^.jr T/ -
'?'! f 1 r
n -j. : -
. -'. { - 1
7 1 . : * -T "'
iimii
nrrrrr
i ;! ! : i-
4. :-;.. ;. i--|.|-.
. . rT,,
1 ' 1 i I"1
f ; '. ' i
!-i ! f M j-
r;-: ;-- r
{ ! M {! ';
trftttf'
vMti-
K
I ::::::::::::: i :::. 5 .;.
^mhtitb
- . r -. . i . . - i [ f t
! i ! i ; ! ! r : "} T*1"
I ~~" :""?*: :' ! J ""'.
Ctrj-pt _
...i.t-tf-t:
.. 1 t *. i« (. )
I 1 i 1 i j !
1NT|r..r,., ,r
- ! .: j.f.j-i t : ! -f-f }.-} f -j |-
1 s-i i I P M-1-r ! ! fi ? !!-!
:.:::!:;;:!::
i-l-H-i-l-t!!- !-:
t rrrrl 1 : \ \ i i
::i«;:r:s[.lr:.i-.--tT(-"-T;:-
PEftl?TRATION. cc/g rtf'''!." -hi
- f »-(- -t-l"
O 4C *O -^ CV O 40 *O *^
W f-4 H rH rH rHOO C?
II
{j fv^ ",i'i
r£. -r :.- T. i.
~r-^f. -.-".
3,000 §
2.000 -H
a
i.ooo 2
900 rj
aoo ti
700 g
400 £
500 0)
400 §
300
(L)
200 -H
nj
S
100 0
90 ;)
80 (J
70
iO
40 1*
" 1
S
o
7 S
s
5 OJ
4
3
VO
0)
1 60
>
113
-------
:-..:.-..- -.. .
4-M--!-i-H-4-!-H4-
.HtH-Hh4-!-fci;ttrtHiH
TrmirrnrFTFTm n
";: : | r. t.vt-ti
4-t-t4-r44-i: M I
M
O
4-1
I
H
(LI
VJ
a
CO
3
CO
H
CU
I
a
o
H
4-1
rt
H
4-J
cu
c
cu
a.
81
H
4-1
CO
s
o
I
CO
s
a
s
60
H
114
-------
APPENDIX C
DOC RATE OF ADSORPTION EXPERIMENTS
Kinetics of DOC adsorption were evaluated in two experiments: run 4 and
run 11. These are shown as follows in Figures C-l and C21 The results of
these and an average of the two are discussed in the text.
115
-------
13
O
LU
o
CO
o
60
H
n)
O
4-1
cfl
o
o
2
1
(I/6UU) 9NINIVIM3d OOQ
o a)
v£> 4-J
O 0)
U-l 4-1
CO
C ct)
o &
H
4-J i-H
O O
CD i-l
O O
o &
CO 0)
CJ 4J
rH cd
4-1 >
-------
i
o
o
DC
O
O
A O D <1 > O
r
ro
or
o
UJ
IE
P
( I
00 ' *° * CVJ O
9NlNIVIAI3d OOQ
CN
cu
S-i
3
«xO
H
Pn
117
-------
APPENDIX D
FREUNDLICH ISOTHERM COEFFICIENTS
The Freundlich isotherm coefficients were calculated as follows. As can
be seen, the 95% confidence interval in most cases included unity (1.0), indi-
cating that a linear isotherm would be an acceptable model for the data.
TABLE D-l. FREUNDLICH ISOTHERM COEFFICIENTS
Subtracting
C
Run Carbon
Number Type Adsorbable Kp
n
as Non-
n
95% Confidence
Interval of
b = 1/n
Points
Included Number
(All with of
Ce > x) Points
Run 6 F400
0.73
6.57 1.2531 0.7980±0.1470 0.9345 C,
0.98
12
Run 8 F400 1.00 8.95 0.8737 1.1445±0.8129 0.7036
60M 1.00 8.85 0.7221 1.3848±0.8275 0.8223
42M 1.00 4.27 0.8076 1.2383*0.5926 0.8028
30M 1.00 6.40 1.8560 0.5388*0.5430 0.4786
Ce > 1.87
Ce > 2.43
(all)
(all)
7
6
8
8
118
_
-------
APPENDIX E
LINEAR ISOTHERMS FOR F400 FOR VARIOUS EXPERIMENTS
Two linear isotherm experiments were conducted on F400 (runs 6 and 8);
these were conducted at two different wastewaters, which had different concen-
trations of DOC. The concentrations of the controls for these are included in
Figure E-l. In run 6, a higher control concentration existed than for run 8,
and correspondingly, the equilibrium concentrations (Ce) were higher for a
given q in run 6 than in run 8.
Further, the equilibrium data for various kinetic experiments with F400
are also included. These are discussed in the text.
119
-------
00
o
"2
8 80
0>
o
>
^ 60
o
o
TO
0)
S 40
T3
O
O
8
20
-------
APPENDIX F
ROOTS OF tan qn IN ANALYTIC SOLUTION TO DIFFUSIVITY
The roots of tan qn for the expression
tail qn = 3qn/(3 + a qn2)
which are essential in solving Eq. 21 are as follows:
TABLE F-l. ROOTS OF tan qfl = 3qn/(3 + aqn2) (Ref. 68)
Fractional
Uptake
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
a
oo
9.0000
4.0000
2.3333
1.5000
1.0000
0.6667
0.4286
0.2500
0.1111
0
<1
. 3.1416
3.2410
3.3485
3.4650
3.5909
3.7264
3.8711
4.0236
4.1811
4.3395
4.4934
q2
6.2832
6.3353
6.3979
6.4736
6.5665
6.6814
6.8246
7.0019
7.2169
7.4645
7.7253
13
9.4248
9.4599
9.5029
9.5567
9.6255
9.7156
9.8369
10.0039
10.2355
10.5437
10.9041
14
12.5664
12.5928
12.6254
12.6668
12.7205
, 12.7928
12.8940
13.0424
13.2689 '
13.6133
14.0662
15
15.7080
15.7292
15.7554
15.7888
15.8326
15.8924
15.9779
16.1082
16.3211
16.6831
17.2208
qfi
18.8496
18.8671
18.8891
18.9172
18.9541
19.0048
19.0784
19.1932
19.3898
19.7564
20.3713
121
-------
APPENDIX G
COMPUTATION OF DIFFUSION COEFFICIENTS
TABLE G-l. COMPUTATION OF DIFFUSION COEFFICIENTS FOR 60M, 42M, 30M, F400
Time
1 min.
2 min.
5 min.
10 min.
20 min.
1 hr.
3.5 hr.
24 hr.
Time
min.
2 rain.
5 min.
10 min.
20 min.
1 hr.
3.5 hr.
24 hr.
60M
De (m2/sec)
2.9 x 10~10
2.3 x 10~10
1.8 x 10~10
2.1 x 10~10
4.9 x 10~10
8.2 x 10 J°
3.3 x 10~10
42M
D£ (m2/sec)
1.99 x 10~J-°
1.64 x 10~J"°
1.64 x 10~10
1.13 x 10 X°
2.36 x 10~~10
1.47 x 10 10
1.45 x 10~10
F = 0.8827
T (unitless)
2.08 x 10~4
3.24 x 10~;
6.36 x 10~4
1.50 x 10 3
6.94 x 10 3
3.47 x 10 2
4.86 x 10~2
F = 0.8421
T (unitless)
2.65 x 10~4
4.34 x 10 4
1.08 x 10~3
1.52 x 10 3
6.26 x 10 3
1.17 x 10~2
4.05 x 10 2
R = 94,900
f (calc.) f
(unitless)
0.3141
0.3692
0.4626
0.5902
0.7988
0.9392
0.9572
R = 50,200
f (calc.) f
(unitless)
0.2784
0.3364
0.4621
0.5125
0.7258
0.8082
0.9295
(experiment)
(unitless)
0.3180
0.3702
0.4547
0.5876
0.7963
0.9329
0.9565
1.0000
(experiment)
(unitless)
0.2734
0.3359
0.4635
0.5117
0.7279
0.8073
0.9297
1.0000
TABLE G-l cont.
122
-------
TABLE G-l cont.
30M
F = 0.7555
R = 18,300
Time
1 min.
2 min.
5 min.
10 min.
20 min.
1 hr.
3.5 hr.
24 hr.
Time
1 min.
2 min.
5 min.
10 min.
20 min.
1 hr.
3.5 hr.
24 hr.
De (m2/sec)
0.38 x 10~10
0.97 x 10 10
0.97 x 10 10
0.97 x 10"10
1.1 x 10 10
0.65 x 10~10
0.44 x 10 10
F400 (Runs
De (m2/sec)
2.5 x 10~10
1.3 x 10 10
1.3 x 10~10
1.3 x 10~J-°
1.3 x 10~10
1.1 x 10~}°
0.67 x 10~10
T (unitless)
1.38 x 10~3
7.10 x 10~4
1.77 x 10~3
3.55 x 10~3
8.11 x 10~3
1.42 x 10~2
3.31 x 10~2
4, 11) F =
T (unitless)
2.92 x 10~4
3.02 x 10 4
9.73 x 10~4
1.53 x 10 3
3.02 x 10 3
8.08 x 10 3
1.64 x 10 2
f (calc.)
(unitless)
0.3798
0.2921
0.4120
0.5178
0.6517
0.7401
0.8576
0.8476 * R =
f (calc.)
(unitless)
0.2971
0.3009
0.4560
0.5232
0.6275
0.7683
0.8522
f (experiment)
(unitless)
0.3788
0.2888
0.4093
0.5109
0.6488
0.7460
0.8505
1.0000
60,300
f (experiment)
(unitless)
0.2950
0.3014
0.4580
0.5226
0.6274
0.7684
0.8525
1.0000
( TABLE G-2. COMPUTATION FOR PORE DIFFUSION MODEL BASED ON
AN ASSUMED MEDIAN D FOR
F400
Time
1 min.
2 min.
5 min.
10 min.
20 min.
1 hr.
3.5 hr.
24 hr.
(Run 4, 11) F =
D (m2/sec)
1.29 x 10~10
0.8476 R =
T (unitless)
1.52 x 10~4
3.03 x 10~4
7.58 x 10~4
1.52 x 10~3
3.03 x 10 3
9.10 x 10~3
3.18 x 10 2
2.18 x 10"1
60M, 42M, 30M,
F400
56,700 Median D = 1.29 x 10~10
f (calc.)
(unitless)
0.2283
0.3015
0.4201
0.5223
0.6282
0.7837
0.9143
0.9942
f (experiment)
(unitless)
0.2950 . . '
0.3014
0.4580
0.5226
0.6274
0.7684
0.8525
1.0000
TABLE G-2 cont.
123
-------
TABLE G-2 cont.
60M F = 0.8827
R = 94,900
Median D = 3.96 x 10
-10
Time
1 min.
2 min.
5 min.
10 min.
20 min.
1 hr.
3.5 hr.
24 hr.
Time
1 min.
2 min.
5 min.
10 min.
20 min.
1 hr.
3.5 hr.
24 hr.
Time
1 min.
2 min.
5 min.
10 min.
20 min.
1 hr.
3.5 hr.
24 hr.
o
D (m /sec) T (unitless)
3.96 x 10~10 2.08 x 10~4
4.16 x 10 4
1.04 x 10 3
2.08 x 10~3
4.16 x 10 3
1.25 x 10~2
4.37 x 10~2
2.99 x 10"1
42M F = 0.8421 R = 50,200
0
D (m /sec) T (unitless)
1.64 x 10~10 2.17 x 10~4
4.35 x 10~4
1.09 x 10 3
2.17 x 10 3
4.35 x 10~3
1.30 x 10 2
4.56 x 10 2
3.13 x 10 1
30M F = 0.7555 R = 18,300
D (m2/sec) T (unitless)
0.97 x 10~10 3.54 x 10~4
7.08 x 10 4
1.77 x 10~3
3.54 x 10"3
7.08 x 10~3
2.13 x 10~2
7.43 x 10 2
5.09 x 10 1
f (calc.)
(unitless)
0.3168
0.4060
0.5387
0.6418
0.7384
0.8631
0.9532
0.9919
Median D
f (calc.)
(unitless)
0.2571
0.3366
0.4624
0.5674
0.6729
0.8211
0.9380
0.9903
Median D
f (calc.)
(unitless)
0.2215
0.2945
0.4150
0.5211
0.6336
0.8025
0.9429
0.9560
f (experiment)
(unitless)
0.3180
0.3702
0.4547
0.5876
0.7963
0.9329
0.9565
1.0000
= 1.64 x 10~10
f (experiment)
(unitless)
0.2734
0.3359
0.4635
0.5117
0.7279
0.8073
0.9297
1.0000
= 0.97 x Kf10
f (experiment)
(unitless)
0.3788
0.2888
0.4093
0.5109
0.6488
0.7460
0.8505
1.0000
124
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-30-123
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
PREPARATION AND EVALUATION OF POWDERED ACTIVATED
CARBON FROM LIGNOCELLULOSIC MATERIALS
5. REPORT DATE
.August 1980 (Issuing Date)
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Paul V. Roberts,, Douglas M. Mackay, and Fred S. Cannon
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Department of Civil Engineering
Stanford University
Stanford, California. 94305
10. PROGRAM ELEMENT NO.
A36B1C
11. CONTRACT/GRANT NO.
Grant No. EPA-R-803188
12. SPONSORING AGENCY NAME AND ADDRESS
Municipal, Environmental Research Laboratory,Cinti,OH
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
13. TYPE.OF REPORT AND PERIOD COVERED
Final Nov. 1976 - Oct. 1979
14. SPONSORING AGENCY CODE
EPA/600/14
15. SUPPLEMENTARY NOTES
Project Officer - Dr. Richard A. Dobbs (513/684-7649)
16. ABSTRACT
This research project was conceived as a preliminary evaluation of the technical
feasibility of converting, solid wastes into adsorbents suitable for wastewater
treatment. The work emphasized the pyrolysis of solid wastes rich in organic
constituents, mainly agricultural wastes. The char prepared from one of these
materials (prune pits) was subsequently activated for comparison with activated
carbons that are widely used in water and wastewater treatment.
The chars so prepared showed specific surface areas of 300 to 650 m^/g,
measured by COg-BET adsorption (195K), but the pores were so small that the
solids were penetrated only slowly by N2- Pyrolysis at 700° to 900°C resulted
in a greater char specific surface than did pyrolysis at 500°C. The activated
carbons made from prune pits demonstrated favorable adsorption performance, when
compared with an activated carbon widely used in water and wastewater treatment.
The prune pit char activated at 60 min. demonstrated a higher adsorption capacity
and superior adsorption kinetics compared to the commercial product (Filtrasorb
400), when judged according to the uptake of dissolved organic carbon (DOC) from
secondary effluent.This difference coincided with a greater surface area and
>macro- and transitional (3 to 300 nm) pore volume for the activated carbon made
from prune pits. An adsorbent made by activation of prune pit char for 42 min.
was approximately equivalent to Filtrasorb 400 in every respect.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
o. COSATI Field/Group
Sewage treatment; Chemical Removal;
Activated Carbon Treatment
Physical Chemical
Treatment
13B
13. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
139
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
125
if U.S. GOVERNMENT PRINTING OFFICE: 1980657-165/0119
-------
------- |