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 prune—pit 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
400—1.4 x 10    m /s—fell 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 "ink—bottle" 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 so—called 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 ) three—dimensional 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 45—200 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 5—7% 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 (5—inch 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
water—cooled 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,
         extrac€ives7~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 fire—retardant 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.  Ash—free 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 ash—free 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
transitional—pores 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
 transitional—and 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 Reuse—From 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 Adsorbers—Equilibrium  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
  o—o 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 C—21   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:  1980—657-165/0119

-------

-------