EPA-650/2-74-084
SEPTEMBER 1974
Environmental Protection Technology Series













-------
                              EPA-650/2-74-084
    ODOR  REMOVAL
         FROM  AIR
    BY  ADSORPTION
      ON  CHARCOAL
               by

      B.C. Kyle and N. D. Eckhoff

       Kansas State University
       Manhattan, Kansas 66502
        Contract No. EHSD 71-4
         ROAP No. 21AXM-061
      Program Element No. 1AB015
   EPA Project Officer: Belur N. Murthy

      Control Systems Laboratory
  National Environmental Research Center
Research Triangle Park, North Carolina 27711
           Prepared for

 OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
      WASHINGTON, D.C.  20460

          September 1974

-------
This report has been reviewed by the Environmental Protection Agency
and approved for publication.  Approval does no"t signify that the
contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
                                 11

-------
ODOR REMOVAL FROM AIR BY ADSORPTION
             ON CHARCOAL
                 by
          Benjamin G. Kyle
 Department of Chemical Engineering

                 and

           N. Dean Eckhoff
  Department of Nuclear Engineering
       Kansas State University
       Manhattan, Kansas 66506
           Project Officer

          Dr. Belur Murthy
     Division of Control Systems
          Technical Center
  Research Triangle Park  NC   27709

-------
                                ABSTRACT
     The efficacy of charcoal for removing odorous organic vapors from
extremely low concentrations in air has been evaluated at ambient conditions.
Specifically, two systems were studied in detail:  Acetaldehyde - Pittsburgh
PCS charcoal, and Ethylmercaptan - Pittsburgh BPL charcoal.
     Fixed-bed breakthrough data were taken for the Acetaldehyde - Pittsburgh
PCS charcoal system at an Acetaldehyde concentration of 28 PPM.  These data
were analyzed by means of a modeling approach which indicated that the
rate-controlling step was intraparticle diffusion.  Preliminary design
calculations based on this work indicate that fixed-bed adsorption with
charcoal is a feasible process for the 98% removal of small concentrations
of acetaldehyde from polluted air.
     An integral reactor approach using Pittsburgh BPL charcoal was utilized
in the study of the catalytic oxidation of ethylmercaptan in the concentra-
tion range 20 to 125 PPM.  Diethyldisulfide was identified as a reaction
product and the reaction was found to be one-half order in ethylmercaptan
concentration.  Preliminary design calculations show that 99% removal of
ethylmercaptan from air by catalytic oxidation on charcoal appears to be
a feasible process.
     This report was submitted in fulfillment of contract number EHSD 71-4
by Kansas State University under the sponsorship of the Environmental Pro-
tection Agency.  Work was completed as of 31 December 1973.
                                    ii

-------
                            TABLE OF CONTENTS







CONCLUSIONS 	   1




RECOMMENDATIONS 	   2




INTRODUCTION  	   3




     Background, Purpose and Scope of Project 	   3




     Description of Project Phases  	   4




     Theory 	   6




EXPERIMENTAL PHASE  	 16




     Differential-Bed Adsorption Studies Using Radioactive Tracers  . . 16




     Fixed-Bed Adsorption Studies 	 27




     Fixed-Bed Catalytic Studies  	 35




DISCUSSION	38




     Acetaldehyde Adsorption Study  	 38




     Fixed-Bed Catalytic Study  	 63




APPLICATION TO DESIGN 	 77




     Acetaldehyde Removal 	 77




     Ethylmercaptan Removal 	 79




REFERENCES	82




NOMENCLATURE  	 83




APPENDIX	85
                                    iii

-------
                             LIST OF FIGURES


Figure                                                                Page

  1.   Charcoal Adsorption System ..................  17

  2.   Charcoal Sample Holder ....................  IB

  3.   1st Test with Ar 41  .....................  20

  4.   1st Test with Ar 41  .....................  21

  5.   2nd Test with Ar 41  .....................  22

  6.   2nd Test with Ar 41  .....................  23

  7.   Test with Ethylmercaptan ...................  25

  8.   Ethylmercaptan Presaturation Test  ..............  28

  9.   Ethylmercaptan Presaturation Test  ..............  29

 10.   Experimental Adsorption System ................  31

 11.   Diffusion Tube ........................  32

 12.   Gas Mixer  ..........................  34

 13.   Static Equilibrium Data for Acetaldehyde on Pittsburgh
       PCB Charcoal at 30°C .....................  41
 14.   Equilibrium Data for Acetaldehyde on Pittsburgh PCB
       Charcoal in Linear Form
 15.   Test of the Freundlich Isotherm Equation ...........

 16.   Pressure Drop Through Beds of Charcoal ............  ^

 17.   Relation Between Mean Residence Time and Effective
       Distribution Ratio for Particle-Phase Diffusion Model  ....  53

 18.   Relation Between Mean Residence Time and Effective
       Distribution Ratio for Pore Diffusion Model
                                                                       54
 19.   Relation Between Mean Residence Time and Effective
       Distribution Ratio for External Diffusion Model   	  55

 20.   Relation Between Mean Residence Time and Effective
       Distribution Ratio for Reaction Kinetics Model 	  56
                                     iv

-------
                       LIST OF FIGURES  (Continued)
 Figure                                                                Page

  21.   Number of Mass Transfer Units versus Residence Time
       for Particle-Phase Diffusion Model  ..............  57

  22.   Number of Mass Transfer Units versus Residence Time
       for Pore Diffusion Model ...................  58

  23.   Number of Mass Transfer Units versus Residence Time
       for External Diffusion Model .................  59

  24.   Number of Mass Transfer Units versus Residence Time
       for Reaction Kinetics Model  .................  60

  25.   Comparison of Particle-Phase Modeling Break Times with
       Experimental Values  .....................  62
 26.   Comparison of Pore Diffusion Modeling Break Time with
       Experimental Values  .....................   67

 27.   Comparison of External Diffusion Modeling Break Times
       with Experimental Values ...................   68

 28.   Comparison of Reaction Kinetics Model Break Time Data
       with Experimental Values ...................   69

 29.   A Plot of Conversion versus W/F  ...............   71

 30.   Typical Effluent History Curve ................   73

 31.   Test of 0.5 Order Reaction ..................   7*


A-l.   Fixed-Bed Test APC-14  ....................   86

A-2.   Fixed-Bed Test APC-18  ....................   87

A-3.   Fixed-Bed Test APC-19  ....................   88

A-4.   Fixed-Bed Test APC-20  ....................   89

A-5.   Fixed-Bed Test APC-21  ....................   90

A-6.   Fixed-Bed Test APC-22  ....................   91

A-7.   Fixed-Bed Test APC-26  ....................   92

A-8.   Fixed-Bed Test APC-29  ....................   93
                                     v

-------
                       LIST OF FIGURES (Continued)







Figure                                                                Paee




A-9.   Fixed-Bed Test APC-30	94




A-10.  Fixed-Bed Test APC-31	95




A-ll.  Fixed-Bed Test APC-32	96




A-12.  Fixed-Bed Test APC-38	97




A-13.  Fixed-Bed Test APC-39	98




A-14.  Gas Chromatograph Calibration Curve for Ethylmercaptan  ....  99




A-15.  Chromatographic Calibration Curve for Acetaldehyde  	 100
                                     vi

-------
                             LIST OF TABLES
Table                                                                 Page

  1.   Results of Fixed-Bed Tests with Acetaldehyde Adsorbed
       on Pittsburgh PCB Charcoal	39

  2.   Results of Static Equilibrium Tests with Acetaldehyde
       and Pittsburgh PCB Charcoal at 30°C	40

  3.   Parameters Obtained from Model Equation for Particle-
       Phase Diffusion Model  	  49

  4.   Parameters Obtained from Model Equation for Pore
       Diffusion Model  	  50

  5.   Parameters Obtained from External Diffusion Model
       Equation	51

  6.   Parameters Obtained from Reaction Kinetics Model
       Equation	52

  7.   Comparison of Calculated N     Values with Modeling
       Values	P?r?	64

  8.   Results of Calculated Nf Values Compared with the
       Modeling Nf	65

  9.   Results of Fixed-Bed Tests with Ethylmercaptan and
       Pittsburgh BPL Charcoal  	  66

 10.   Elements Present in Charcoal Samples as Determined by
       Neutron Activation Analysis  	  76

 11.   Amount of Charcoal Required to Remove Various Percentages
       of Ethylmercaptan from Various Concentrations in Air at
       30°C	80
                                    vii

-------
                               CONCLUSIONS
ACETALDEHYDE REMOVAL
1.   Preliminary design calculations show that fixed-bed adsorption appears
     to be a feasible process for removing up to 98% of acetaldehyde from
     air in the 10 to 35 PPM concentration range.
2.   Equilibrium data for the adsorption of acetaldehyde vapor onto Pittsburgh
     PCB charcoal at 30°C can be represented by the Langmuir Isotherm Equation
     over the range 1 to 35 PPM of acetaldehyde in air.
3.   Fixed-bed breakthrough data for the acetaldehyde vapor - Pittsburgh PCB
     charcoal system can be represented with constant - pattern, internal-
     diffusion models.  The parameters in these models behave realistically.
     The rate of adsorption appears to be controlled by intraparticle
     diffusion.
ETHYLMERCAPTAN REMOVAL
1.   Preliminary design calculations show that 99% removal of ethylmercaptan
     from air in the 20 to 125 PPM concentration range by catalytic oxi-
     dation on charcoal appears to be a feasible process.
2.   Diethyldisulfide has been identified as a product of the catalytic
     oxidation of ethylmercaptan with air during adsorption on Pittsburgh
     BPL charcoal.
3.   The catalytic reaction was found to be one-half order in ethylmercaptan
     concentration.

-------
                             RECOMMENDATIONS
ABSORPTION
1.   It is recommended that the study involving adsorption of Acetaldehyde
     on Pittsburg PCS charcoal be expanded to include concentrations of
     Acetaldehyde as high as 10 mole percent.  The aim of such a study
     would be to determine whether the rate-controlling processes are the
     same for all concentration levels or change as extremely low levels
     are reached.  If the former condition is obtained, it might be possible
     to extrapolate the information available in the literature, the bulk
     of which applies to much higher concentration levels.
2.   It is recommended that the adsorption equilibrium and kinetic behavior
     of other members of the aldehyde family be studied.  Such a study
     could lead to systematic parameter correlations which would allow
     the adsorption behavior of any member of a family of compounds to
     be predicted from the parameters of a few members.
3.   In order to provide all of the information needed for design of
     adsorption systems the regeneration behavior of charcoal, especially
     at low levels of adsorbate loading, should be studied.
MERCAPTAN CATALYTIC OXIDATION
     It is recommended that the study begun here be expanded to include
other members of the mercaptan family and other types of charcoal.  The
effect of temperature and the long term stability of the catalyst should
also be investigated.  Identification of the reaction products should
also be a major part of this study.

-------
                              INTRODUCTION
BACKGROUND, PURPOSE, AND SCOPE OF THE PROJECT
Background
     The Division of Process Control Engineering (now Control Systems
Laboratory) started the odor control project in 1968.  A survey of the
existing control technology for odors revealed the urgent need for the
development of suitable adsorption systems as one of the control devices.
The present plans for odor control research call for detailed investigations
of odor adsorption mechanisms using known odorants and activated charcoal
and/or other suitable adsorbents.  This contract is to develop useful infor-
mation on the design data needed for the construction of efficient odor control
adsorption units.
Purpose
     The contract is expected to accomplish the following:  '(a) development
of suitable experimental techniques so that reliable adsorption equilibrium
and kinetic data may be obtained,  (b) provide data for several homologous
series of adsorbates, (c) from data gathered and analyzed, acquire an under-
standing of the thermodynamics and kinetics of adsorption at low adsorbate
concentrations.  The data and conclusions resulting from these studies will
be used in the scale-up and development of package adsorption devices for
industrial odor control.
Scope
     The scope of the original contract consisted of the following:   (a) Con-
duct a literature survey on the existing data on adsorption equilibrium and
kinetics relating to odors and activated charcoal,  (b) Standardize suitable
analytical techniques like application of radioactive tracers and ionization
detectors for the system selected  for study,  (c) Derive mathematical  expres-
sions relating adsorption efficiencies and operating process variables from
adsorption equilibrium and kinetic data collected in the laboratory,  and
 (d) Conclusions should include design data for scale-up, as well as recommenda-
 tions for further R & D in the field of Adsorption Mechanisms.

-------
     Upon extension of the contract the following items were added to the
scope of the project:  (a) Modify the analytical equipment to increase the
capability for analyzing all the-components of the reaction mixture involved
in the adsorption/desorption of mercaptans on activated charcoal, (b) Con-
struct a differential or integral reactor and adsorption system suitable
for determining the reaction mechanisms and scale-up parameters, (c) Expand
the literature survey effort to include the related studies of catalysis and
regeneration of adsorbents, (d) Conduct fixed bed studies with more than one
type of activated charcoal to test the catalytic ability of other types of
charcoal than the one being presently studied under the existing contract,
(e) Identify and analyze all the reaction products involved in the absorption/
desorption of mercaptans on activated charcoal, and (f) Study the reaction
mechanism using the differential or integral reactor constructed and develop
suitable design equations from the data, for the construction of reliable and
low cost regenerable adsorption control devices for odor control.
DESCRIPTION OF PROJECT PHASES
Differential-Bed Adsorption Studies Using Radioactive Tracers
     The intent of this study was to follow the adsorption kinetics using an
experimental method which consisted essentially of recirculating a gas con-
taining radioactivity-tagged adsorbate through a shallow bed of adsorbent
while continuously monitoring the gas-phase concentration of the adsorbate
with a radiation detector.  A material balance would allow the adsorbent
loading to be calculated for any adsorbate concentration, thus permitting the
determination of instantaneous values of gas-phase adsorbate concentration and
corresponding values of adsorbent loading.  The instantaneous rate of adsorp-
tion could then be determined from the time dependence of the adsorbent
loading.  This then would provide the rate as a function of gas-phase adsorbate
concentration and is the major advantage to the differential bed method.
Unfortunately, this experimental technique could not be successfully imple-
mented because of adsorption of the radioactive component on the radiation
detector.  An integral or fixed-bed approach was then adopted.
Fixed-Bed Adsorption Studies
     This experimental technique involves passing a gas stream  containing a
known and constant concentration of adsorbate through a bed packed with

-------
adsorbent.  Initially, the adsorbent is fresh and practically all of the
adsorbate is adsorbed so that the effluent stream from the bed contains very
little adsorbate.  With the passage of more fluid the adsorbent in the bed
begins to saturate thus causing the concentration of adsorbate in the
effluent stream to rise.  A plot of effluent adsorbate concentration versus
time, or volume of fluid processed, is called a breakthrough curve.  The
shape and position of the breakthrough curve are dependent on the adsorption
kinetics, however, because of the complexity of the fixed-bed process, it is
necessary to employ a modeling approach which only allows the kinetic be-
havior to be determined indirectly.
     The technique of modeling as applied to fixed-bed operations seeks to
simplify the mathematical description of these highly complex systems by
approximating the actual flow patterns with idealized representations and by
specifying a rate-controlling step for the transport of mass between the
fluid and solid phase.  A solution of these simplified descriptive equations,
referred to here as the model equation, takes the form of an expression
relating concentration to time and position within the bed.  The results
predicted by the model equation are then compared with experiment and the
validity of the model is judged by its ability to fit the experimental break-
through data.  The nature of the flow patterns and the rate-controlling mass
transport mechanism are thus deduced by determining which of several models
best fits the experimental results.  Each model will contain one or two
parameters which can be evaluated by fitting the model equation to the
experimental data.  A further test of the model is whether  these parameters
are physically realistic; with a realistic model it is possible to correlate
these model parameters with the system variables so that behavior of similar
systems can be predicted.
Fixed-Bed Catalytic Studies
     The  experimental apparatus and technique for the study of solid-fluid
catalytic reactions in  fixed-beds are similar to those for  fixed-bed adsorp-
tion studies, but  these studies have  the advantage that because the catalytic
fixed-bed system operates at steady state the data are easier  to analyze.
An influent stream containing a known and constant concentration of reactant
is passed at a constant flow rate  through a  packed bed of catalyst particles.
The effluent stream from  this bed  is monitored and when steady state  is
attained  the concentrations of  reactants and products are determined.  Thus

-------
for a single run the conversion is determined for a given ratio of catalyst
weight to influent flow rate.  A differential material balance relationship
allows the reaction rate to be determined from the slope of a curve obtained
from plotting conversion versus this ratio.  It is therefore required that
enough runs be made so that this curve can be established.  Alternately, one
may assume a rate expression, integrate the differential material balance
relationship with this rate expression, and obtain an algebraic equation re-
lating conversion to the catalyst to feed ratio.  This corresponds to the
modeling approach described earlier as each rate expression yields a different
integrated rate equation.  As before, the validity of a rate expression is
determined by how well its corresponding integrated rate equation fits the
conversion versus catalyst to feed ratio data.
THEORY
Fixed-Bed Adsorption
     The development of model breakthrough equations presented here closely
follows that of Vermeulen (1) and begins with a material balance applied over
an infinitesimally small time increment to an infinitesimally thin cross
sectional slice of the bed containing dv volume of adsorbate.  As the fluid
moves through this slice of adsorbent bed the concentration of adsorbate in
the fluid decreases an infinitesimally small quantity dC and the adsorbent
loading in the slice increases by dq.  Therefore, from the principle of con-
servation of mass it follows that
          Amount of adsorbate _ Amount of adsorbate
          entering the slice    picked up by adsorbent
                 Amount of adsorbate   Accumulation of adsorbate
                 leaving the slice     within the void spaces
Written in mathematical language the material balance equation is
                        fac]      faql     fac)
                       - avi, = pb 3v   + e[avj
                        \  jy     \  Jv    \  I
Because C is a function of v and V, it can also be expressed as a function
of v and V - ve.  This change of variable produces the following simplified
equation
                                  -ve

-------
By introducing the following dimensionless concentration variables
                                x = c/co
                                Y = q/q*
and a dimensionless through-put parameter
                             C (V - ve)
                         _    o
                                 pb V
        V - ve
         ve D
equation 2 can be written
                                       3Y
                                       3Zv
(3)
(A)


(5)


(6)
This is the material balance partial differential equation for a fixed-bed
and is the starting point for most model breakthrough equations.
     It is an experimental fact that for many systems the breakthrough
curves retain a constant shape regardless of bed length.  This is because
of a constant rate of movement of all parts of  the exchange zone.  This  can
be stated mathematically as
                             ^..1
                                                                       (7)
                             3v
= constant
or in terms of the through-put parameter
                             3 Zv
                             3v
                                     constant
                                    (8)
The right hand side of the material balance  equation  can be  restated  so  that
equation 6 can be written as follows
                                      3X
                                     3 Zv
          3Y
          3X
rearrangement of equation  9 gives
                           (3X/3v)
                                  Zv
                           (3X/3Zv)
     3 Zv
      3v
                                               =  constant
Equation  10  integrates  to
                          Y =  (constant)X +
                                                                       (9)
                                    (10)
                                    (ID
where C..  is  the  constant  of  integration.   Because  X and Y  are dimensionless
variables and  have  the  same  limits  equation 11  simplifies  to
                                  Y =  X                               (12)
which also leads to
                                  dY =  dX                              (13)

-------
Thus for the constant pattern case the material balance relationship results
in the simple statements of equations 12 and 13.  We will now develop break-
through equations for the constant pattern case where different rate-con-
trolling mechanisms obtain.
     Fluid-phase Diffusion Model - For this model we assume that the slowest,
and therefore rate-controlling, step involves the migration of adsorbate
molecules from the bulk of the fluid phase to the surface of the adsorbent
particle.  Customarily, the rate in such situations is expressed in terms
of a mass transfer coefficient and a characteristic concentration difference

                               =     (X - X*)                        (U)
The X* term is the fluid-phase concentration which would obtain if equilibrium
existed at the solid-fluid interface.  In order to obtain this term something
must be said about the equilibrium relationship.  For a great many systems
it has been found that this can be accomplished in the following manner

                             '* - , + '(HOI                         (15)
or

                             ** • i + 5-Di                         (16)
where R is called the separation factor and is usually assumed to be constant
over the  full range of X and Y.  If this equilibrium relationship is used
along with the constant pattern conditions expressed by equations 12 and 13
it is possible after some simplification to write equation 14 as
                          dX = V X(X-1)(R-1)
                          dt   eD  1 +  (R-l) X
It is more convenient to replace the time variable with the  through-put
parameter Z where the two are related by the following
                                     tF
                                                                      (18)
                                     e Dv
 This  substitution results  in
                           dX

                            dZ     f  1  + (R-l)  X
 where Nf  is  the number of  fluid-phase transfer  units  defined  as
                                                                      (20)

-------
In equation 19 variables are separable and integration yields
                  1
                 1-R
                     Hn X -
 R
1-R
in (1-X)  = Nf(Z-l)  + af              (21)
where a  is the sum of constant terms including the  integration  constant  and
can be evaluated by making a material balance for the entire  column  at
saturation
                           rX=l
                                (l-X)dZ = 1                          (22)
                           'Z=0
Equation 21 is the breakthrough equation for the constant-pattern, fluid-
phase diffusion model; it relates the dimensionless  effluent  concentration,
X, to the through-put parameter, Z,  in terms of an equilibrium parameter,
R, and a rate parameter N..
     Solid-phase Diffusion Model - Diffusion arising from a concentration
gradient within the individual adsorbent particles after the  adsorbate
arrives at the solid-fluid interface is called solid-phase diffusion.   This
includes diffusion through a homogeneous, permeable, non-porous  solid,  and
diffusion in a mobile, adsorbed phase covering the pore surface  of  a porous
solid whose crystalline portion is impermeable.  This is assumed to  be  the
rate-controlling step and by analogy with the more thoroughly studied
phenomenon of fluid-phase diffusion the rate is expressed as
                           ,iv   k., a
                           £ = -ir
-------
                 1-R
                     in X -
                             1-R
                                In  (1-X) = N  (Z-l) + a
                                                   (25)
where N  is the number of particle-phase transfer units defined as
       P
                                    k  a v
                               N  --2^=—                           (26)
                                P      F
and a  is a constant which can be evaluated from the separation factor, R.
     p
     Reaction Kinetics Model - Langmuir derived his famous isotherm equation
by saying that the net rate of adsorption is equal to the rate at which
molecules from the gas phase are adsorbed on the surface minus the rate at
which molecules leave the surface.
                         |f - k[p(qm - q) - q/b]                     (27)
Here the adsorption rate is assumed proportional to the partial pressure of
the gas and the amount of surface unoccupied by adsorbed molecules and the
desorption rate is assumed proportional to the number of adsorbed molecules.
At equilibrium the net rate is zero and equation 27 reduces to the familar
Langmuir isotherm equation.  The form of the Langmuir isotherm equation is
such that it corresponds to equation 15 with the separation factor R constant
with a value between  zero and unity.  Therefore, in terms of dimensionless
concentrations equation 27 can be written
                     dY
                     dt
                           1-R
                                 [X(l-Y) - RY(l-X)]
                                                   (28)
We have  also  seen  that  the  rate  can be  expressed  in  terms of  fluid  and  solid
phase  resistances.  Thus  since the rate through each resistance  is  equal we
can write  from  equations  14 and  23
                        . __ (x _
                                              (Y* _ Y)
                                                   (29)
                          eD  v"   " '   eD
As before, the X* and Y* terms can be eliminated by using the separation
factor.  If this is followed by summing the numerators and denominators
one obtains
                           X	             RY
                              n\v   ITA   1J_ /D i \v
                                                                     (30)
dY = R + (l-R)X
dt=       DE(JL +
             VV a
                                             1 + (R-l)Y
                                        1)
                                          j
                                    10

-------
When equations 28 and 30 are equated and the resulting equation simplified
the familar reciprocal form for resistances in series results, but modified
by a correction term B
                              3. _  1  .   1                           /on)
                              k - k a + k,a                          (  ;
                                   P     f
The correction term B is a function of X, Y, and R but has been evaluated
and found to depend mainly on R (1).  Therefore, for a given adsorption
system the correction factor may be regarded as constant.
     Except in rare cases, adsorption at the solid surface is usually found
to be extremely rapid and the resistance due to this step will generally be
negligible.  However, this mechanism finds wide useage because with the
correction term 3 the rate expression is equivalent to that written for
combined particle-phase and fluid-phase resistances.
     The development of the breakthrough equation for the reaction kinetics
model begins with equation 28 and follows the same general procedure used
for the two previously developed models.  The result is
                                       =ND(Z-1)                     (32)
                          1-R
1-X
where N  is the number of reaction units defined as
       R
                                   k p  D ve
                               R    F(l-R)
     Pore Diffusion Model - Pore diffusion refers to the diffusion occurring
within the fluid-filled pores of the solid; the driving force for this
process is the concentration gradient existing in the fluid within the
pores.  If the fluid-phase concentration in the pores is assumed negligible
compared to the solid-phase value, the equation of continuity can be written

                                         ->„ H
In equation 34 the terms c and q represent point values of the fluid and
solid-phase concentrations respectively.  The average solid-phase concentra-
tion is

                            -   3   fTp   2
                            q = AT     qr  dr                        (35)
                                r  J0
                                 P
                                    11

-------
Differentiation of equation 35 gives


                                       2

                                                                     (36)
                           3r idtj    r3

                                       P


Combination of equation 36 with the continuity equation yields




                           H-   3D     '"an
                           Qfl      pOTTS	 o v

                           dt =       3    37




The pore diffusion model has been worked out only for the case of irreversible


adsorption (R = 0).  This allows the assumption that the concentration wave


penetrating the particle will saturate each spherical layer before it


penetrates further.  That is, C cannot exceed zero at a particular radius


r = r. until q at that r has become equal to the saturation value q.  Thus


q = q_ for r. < r < r  and q = 0 for r < r..  This simplifies equation 35  to
     Tn      i        p                    i
                                         •-  •                          <38)
                           Sn
From equation 38 we obtain



                           ,-     3 q_ r.2 dr,

                                                                      (39)
       3 q_ r2 dr.
dq _     in  i 	i
                           dt        r3   dt

                                      P


substitution of  this expression  into equation  37  results  in
                        dt
                                    Pb  ri
          2  3C
The  term r  T—  can now be  evaluated  from  the  original  equation  of  continuity.
             or

If equation  34  is applied  in  the  region r.  <  r  <  r  where,  according  to  the


model, -r^ =  0 we obtain
       o t

                            2  8C
                           r -5- = constant  =  C.,                       (41)
                              or               J.


Equation 41  can be integrated  subject  to  the  following boundary conditions


                       C = C..  = CX    at    r  =  r                      (42)
                            to             p


                               C = 0    at    r  =  ^                    (43)



where  Cf is  the concentration in  the fluid  surrounding the  particle.   The

  2 3C
r —  term  can  now be written
   or
                                     12

-------
                                    CX
                             2 8C    Q                               (44)
                               *r    r  - r.
                                      P    i
When equation 44 is substituted into equation 40 the resulting equation of
continuity becomes
                    dr.      C  D     r        Y
                      i _     o  pore  p       X                     ^^

Because X is the dimensionless concentration in the fluid surrounding the
particle, the constant pattern condition can be applied and X replaced by
Y as given by equation 38.  On simplification, equation 45 can be expressed
in terms of r. and t
2 .
(k awr + r r . H
pore p pi
15 J I ri
60 D CQ
pore ,^
b Tn p
K*

                    dt
where
                                   60 D^ _  C.
                                                                      (47)

Integration  of  equation 46 gives r. as a function of  time.  This can  be
followed by  numerical evaluation of Y via equation 38,  and application of
the  constant pattern condition  (equation 12) yields X as  a function of t
or Z.   The result  fits the empirical relation
        X =  0.557[N    (Z-l)  +  1.15] - 0.0774[N     (Z-l)  + 1.15]2     (48)
                    pore                        pore
where
                                    k     a v
                             N     = _E°If	                         (49)
                              pore       F
The  four model breakthrough  equations developed  here  (equations 21,  25,  32,
 and  48) are  the ones applied to the experimental breakthrough data obtained
 for  this  project.   There  are many  other models found  in the  literature which
 could be  used, but these  four cover  the  complete range  of mechanisms  and
 are  widely used.  Their popularity undoubtedly stems  from their relative
 simplicity and the fact  that the behavior  of  the two  parameters is easily
 predicted qualitatively,  if  not quantitatively.   These  considerations are
 of paramount importance when design  is  the ultimate objective.
                                      13

-------
Fixed-Bed Catalysis
     Because a fixed-bed catalytic reactor operates under steady state
conditions the material balance equation is an ordinary differential equation
rather than the partial differential equation as required for fixed-bed
adsorption.  If we focus our attention on a very thin cross-sectional slice
of the catalyst bed which contains dW mass of catalyst a material balance
for the reactant is
                         F. dx. = VC. dx. = r dW                     (50)
                          Ao  A     Ao  A
The reaction rate based on a unit mass of catalyst, r, is expected to be a
function of temperature and the various reactant and product concentrations,
however here we will restrict our attention to isothermal operation.
     There are two ways in which equation 50 can be applied to the analysis
of data from a fixed-bed catalytic reactor.
     Equation 50 can be applied in differential form.  Rearrangement gives
                           dx.     C.  dx.
                             A      Ao   A = r                        (51)
                          d(W/F)   d(W/V)
Thus, it is seen  that  the rate can be obtained from the slope of a curve of
x. versus W/F or  W/V.  The customary practice is  to operate the reactor
 A
under various conditions and make a plot of XA vs W/F or W/V for a constant
C  ; the slope  of this curve gives the rate at that particular conversion
 Ao
or reactant concentration.  The concentration dependence of the rate  is then
determined by examining these data.
     The alternate approach is to substitute various rate equations into
equation 50 and integrate to obtain an expression involving XA and W/F or
W/V.  Various curve fitting procedures are then used to test this integrated
equation against  the experimental data.  The test of these integrated
equations  lies  in how well  they are able to fit the data.  As an example of
this approach,  consider a 1/2 order and irreversible mechanism.  The  rate  is
                            r = kC.  (l-x.)1/2                         (52)
                                  Ao    A
When  this  expression is substituted into equation 50 and integration  is
performed  with  the boundary condition  that XA = ° when W/F = °  the following
integrated rate equation  results

-------
                             2C1/2
                       w/v = _Ao_ tl _ (l-xA)i/Z]                   (53)

                                                                     1/2
Thus, this proposed mechanism could be tested by plotting [1 - U-XA)   J

versus W/V.  If a straight line passing through the origin is obtained, the

mechanism is said to be valid.
                                      15

-------
                            EXPERIMENTAL PHASE
 DIFFERENTIAL-BED ADSORPTION STUDIES  USING RADIOACTIVE TRACERS
 Experimental Apparatus
      The volumetric adsorption  apparatus used  radioactive  tracers  for
 measurement  of  the  extremely low concentration of  the adsorbate  in the gas
 phase.   A gas mixture consisting of  air and  the tagged adsorbate was circu-
 lated through a shallow bed of  charcoal while  the  gas phase was  continuously
 monitored with  a radiation  detector.  The detector output  allowed  the instan-
 taneous  partial pressure of the adsorbate to be determined.  If  the assumption
 of perfect gas  mixture  behavior is made,  a material balance allows  the
 instantaneous asorbent  loading  to be calculated
                                  VP_  *  P)
                              q  - —IS	                          <54>
      The instantaneous  values of p and  q  allow a study of  the adsorption
 kinetics to  be  made.  When  a constant value of adsorbate partial pressure
 is attained  the equilibrium values of p and q  can  be determined from
 equation 54.
      A schematic diagram of the proposed  adsorption system is shown on
 Figure 1 and a  drawing  of the charcoal  sample  holder is shown in Figure 2.
 Gas flow could  be routed through  the charcoal  sample holder or bypassed by
 proper positioning of the 4-way stopcock.  The  charcoal sample holder, cold
 trap, and  coil  were fabricated  from pyrex glass.  These pieces were connected
 by Teflon  tubing with the exception of  the piping adjacent to the diaphram
 circulating  pump which  consisted  of copper tubing with brass valves and
 fittings.  The volume of the system,  V  , could be varied depending on the
                                      s
 diameter and  length of pyrex tubing in  the coil.  A differential pressure
 transducer in conjunction with  a pressure gauge was used to measure the
 pressure in  the system to within ± 0.2 mm Hg.  The pressure transducer was
 also used to monitor the flow of gas  through the radiation detector by
measuring the pressure drop.  The entire apparatus, with the exception of
 the radiation detector and  the pressure gauge was contained within a thermo-
 stated air bath.  The radiation detector was a Packard TRI-CARB Scintillation
 Spectrometer initially equiped with a flow cell fabricated from plexiglass
and packed with anthracene crystal scintillators.

                                     16

-------
CHARCOAL ADSORPTION SYSTEM
                 MERCURY
                 MANOMETER
                                                      PRESSURE
                                                      GAUGE
                                                     PRESSURE
                                                     REGULATOR
                                              PRESSURE TRANSDUCER
                                                  INDICATOR
                                          i PRESSURE
                                          J TRANSDUCER
    FIGURE I

-------
Fig. 2.  Charcoal  Sample Holder
 8mm  tubing
                      T  29,
                           '42
                                    2-hooks
                                    fritted  glass
                                     (coarse)
                      4-way stopcock
                o-ring  joint
                     18

-------
     The following operating procedure was followed:
     1.  Using an analytical balance the desired quantity of charcoal (about
10 to 20 grams) was weighed into the charcoal sample  holder.
     2.  With the charcoal sample holder stopcock in  the bypass  position the
air pressure in the system was reduced to [760-p ]  mm Hg.  The air was then
circulated through the cold trap containing charcoal  at a lower  temperature
so that any adsorbable components were removed.
     3.  With the cold trap out of the system the gaseous adsorbate was
admitted to the system through the intake valve until the pressure in the
system reached 760 mm Hg.  Pressure readings before and after introduction
of the adsorbate were made,  (p  was in the range of  20 to 80 mm Hg.)
     4.  The gas mixture was then circulated through  the system  until a
reliable initial reading was obtained from the radiation detector.
     5.  At time zero the charcoal sample holder was  connected into the
system and the output of the radiation detector continuously monitored.
     6.  After equilibrium had been attained the charcoal could  be discarded
or steps 2 through 5 could be repeated in order to obtain data at higher
adsorbent loadings.
Argon 41 Tests
     To test the feasibility of using the Packard TRI-CARB spectrometer system
and  to learn the mixing  characteristics of the experimental adsorption apparatus,
two  tests with the radioactive isotope Argon 41 were performed under different
operating conditions.
     The results of the  two tests are shown in Figures 3, 4, 5,  and 6.
Figures 3 and  5 show the TRI-CARB reading in counts per minute  (CPM) versus
time.  The maximum of these curves indicated the time required  for complete
mixing and the discontinuity following the maximum represents the inclusion
of the charcoal holder into the system.  Figures 4 and 6 are semilog plots
of CPM vs time.  From the  slope of these straight lines  the half life  can be
computed.  These values  of  1.80 hr and 1.95 hr are in excellent agreement with
the  accepted value of 1.83  hr.
     There appeared to be  almost no mixing effect when the  charcoal holder
was  switched  into  the system.  The volumes of the system with and without
the  charcoal  holder were 970cc and 929cc  respectively and on switching the
                                     19

-------
I
 8
 o
     FIGURE  3


     l" TEST  with Ar 41
 §
 o
                     100
200
300
400
                                                                                       Time  Min.
500

-------
                                    FIGURE 4
                                    IST  TEST  WITH Ar4l
                                           HALF LIFE'1.95 IT.
1000
100
            50
100        BO
     TIME  MIN
          21
                                         200
250
300
                                                                      350

-------
  o
  o
f\
a.
U
  o
  o
  o
  •o
                                              FIGURE   5


                                              2"d TEST  with Ar41
               40
                      80
120
160
200
240
280  .     .  320
    time  mm.

-------
o
9000

8000

7000

6000


5000
  4000
  3000
  2000
   1000
                  48
                                  FIGURE  6
                                  2nd  TEST with Ar4l

                                        HALF LIFE'1.8 hr.
                            96
                             TIME  MIN.
                                23
144
192
240

-------
charcoal holder into the system one would expect a dilution of the radio-
activity in the proportion to the ratio of these volumes (i.e., 929/970 = .948)
Readings taken from Figures 3 and 5 gave the following results:
                                        ^   3380
                  First test      -- =      = .952
                (CPM
Second test
                                  (CPM)1   6060
The above results indicated that this mixing effect was negligible and
provided further indication, along with the half life determination, that
reliable data could be obtained from the TRI-CARB system.
Ethylmercaptan Tests
     Having confirmed that the TRI-CARB system and the circulation system were
                                41
performing as expected with the   Ar, the first experiment using ethylmer-
captan was attempted.  Sulfur 35-tagged mercaptan was used for this test.
     After the e thy liner cap tan was added the gas was circulated through the
Packard TRI-CARB flow cell (at this point the cold trap and charcoal holder
are isolated) and readings were taken over a 2 1/2 hour period.  The detector
readings during this period are shown in Figure 7 as region I.  The response
in region I indicates loading (adsorption and or absorption) of anthracene
with ethylmercaptan and was an unexpected event.
     After this circulation period the charcoal holder was switched into
the system and adsorption of ethylmercaptan was observed to occur.  Region
II of Figure 7 shows the detector reading during this period.  Because of
the loading of the anthracene the adsorptive action of the charcoal was
overshadowed at lower concentrations.
     The exact nature of the anthracene loading process is not known although
it was found that the ethylmercaptan was tightly bound to the anthracene.
After the test run the anthracene crystals were removed from the flow cell
and allowed to stand in an open beaker for three days.  A sample of the anthra-
cene was then dissolved in toluene and the level of radioactivity was deter-
mined with a scintillation counter.  This measurement showed that a large
quantity of ethylmercaptan remained and indicated that the "loading" process
is characterized by strong attractive forces.
                                     24

-------
                       FIGURE  7
                TEST with ETHYL  MERCAPTAN
      REGION  I
REGION  2
2 XIO6
 XIO
                           4     5
                           TIME HR.
                           25
                                 10

-------
     When the plexiglass flow cell was removed from the circulation system,
emptied of anthracene, repacked with fresh anthracene, and then replaced
in the TRI-CARB system with no gas flow, a high count rate was observed.
This indicated that ethylinercaptan was adsorbed or absorbed in the walls
of the flow cell.
     In an attempt to eliminate or reduce the loading, a flow cell fabricated
from Kel-F plastic and packed with Lithium glass bead scintillators was
installed.  Tests with   Ar showed this to be a very efficient detection
                            O C
system, however a test with   S-tagged ethylmercaptan again revealed that
loading had occurred.
     Because it appeared that the loading phenomenon could not be eliminated,
efforts were then directed toward controlling it.  This would be attempted
by removing the radioactive mercaptan from the flow cell, the Lithium glass
bead scintillators, and the inside surfaces of the adsorption system and then
presaturating these parts with non-radioactive mercaptan.  Previous tests had
shown  that the mercaptan is very strongly adsorbed and it was believed  that
once the  system had been presaturated, or loaded, that no further loading
would  occur when the radioactive mercaptan was introduced.  Because adsorptive
forces are temperature dependent, it was reasoned that increasing the operating
temperature of the flow cell from the specified value of 5°C  to room temper-
ature  should serve to decrease the extent of  loading on the flow cell.
     Considerable difficulty was encountered  in attempting to remove the
loaded radioactive mercaptan.  Heating  the Lithium glass bead scintillators
under  vacuum for several hours reduced  the amount of loaded mercaptan,  but
when these beads were packed in the flow cell a very high background count
was obtained.  After  some  experimentation, a  sizeable  reduction in background
count  was obtained by replacing the Lithium glass bead scintillators with
new beads, soaking the Kel-F flow cell  in toluene and  keeping the entire
adsorption system under vacuum  (less  than 5 mm Hg) for approximately a  week.
These  rather drastic  measures reduced the background  count to about 1300 cpm
with air circulating  through the adsorption system.   Although this is about
six  times the background count obtained for the original  test of  this flow
cell,  the count  remained constant  for about an hour  and  the decision was
made  to  proceed  with  the presaturation  step.   Non-radioactive mercaptan was
 introduced into  the  system and  the  gas  was circulated for almost  an hour.
The  circulating  gas  was  then routed through  a liquid nitrogen-cooled  trap
                                     26

-------
for several hours in order to remove any unadsorbed non-radioactive mercaptan.
No sign of condensation was observed in the trap indicating that all of the
charged mercaptan was adsorbed on the various parts of the system.   During
this presaturation period the TRI-CARB readings were observed in increase
slightly therefore the measurements were extended for several more  hours.
     The TRI-CARB readings during the presaturation period are shown on
Figure 8.  These readings have been corrected for the background count of
1300 cpm.  The increase in the TRI-CARB output with time prior to intro-
duction of the radioactive mercaptan is difficult to explain.  The only
possible explanation seems to be that all of the radioactive material was
not removed from the interior surfaces of the circulating system during the
cleaning process and that some of this material was displaced by the non-
radioactive mercaptan.  This exchange process would be expected to occur
rather slowly as observed.
     As can be seen from Figure 8, the Tri-CARB readings showed signs of
leveling off after about seventeen hours.  At that time a test with radio-
active ethylmercaptan was started using the previously described operating
procedure.  The TRI-CARB output during this test is shown on Figure 8 and on
a different time scale on Figure 9.  Figure 9 shows that a state of equili-
brium appears to occur approximately 45 minutes after the charcoal is
switched into the system.
     Because of  the apparent exchange of radioactive and non-radioactive
mercaptan  presaturation of  the system was not considered feasible.  Since
the loading phenomenon could be neither eliminated nor controlled  it was
decided  to abandon  this experimental approach and  investigate  the  possibility
of using other experimental approaches.
FIXED-BED  ADSORPTION  STUDIES
     The  fixed-bed  approach was chosen because  it  is a well  established
 technique  and would allow kinetic  and  equilibrium  data  to be determined  under
 essentially  the  same  conditions as prevail in  industrial adsorbers.  The
method  consists  of  passing a fluid stream  of a  constant composition at a
 constant flow rate  through a fixed bed  of  adsorbent which  is initially free
 of the adsorbate.   The effluent  fluid  concentration is monitored and  the
 effluent concentration versus time data is referred to  as  breakthrough data.
 A plot of  this  data is referred  to as  the  breakthrough  curve.
                                     27

-------
CPM
FIGURE  8.   TEST  WITH  ETHYLMERCAPTAN
                                                                                CHARCOAL IN
15000
10000
 NJ
 CO
 5000
                                    NON-RADIOACTIVE MERCAPTAN SATURATK
                                              10
                                                                  15
                                                 TIME (hr.)
                                                                               RADIO ACTIVE MERCAPTAN IN
                                                                     20

-------
                      FIGURE  9.  TEST    WITH   ETHYLMERCAPTAN
20000
15000
  10000
  5000
                                      Charcoal In
            Rado Active Mercaptan In
                              0.25
0.5
                                    TIME (hr)
                                              29
0.75

-------
     The capacity of the adsorbent can be obtained by a material balance
calculation involving integration of the area above the breakthrough curve.
This adsorbent capacity is usually the same as the capacity of the adsorbent
when in equilibrium with the influent fluid and is customarily taken to be
the equilibrium capacity.
     A modeling approach is used for a kinetic analysis of fixed-bed
adsorption.  Here a model is defined as a mathematical expression arrived
at through specifying an equilibrium relationship, a rate expression, and
appropriate boundary conditions.  Many models are possible and the mathe-
matics have been worked out and published for quite a number.  In applying
the modeling approach one attempts to fit the experimental breakthrough data
to a selected model breakthrough equation.  This process of curve fitting
also permits the evaluation of the model parameters.  The validity of a given
model is then determined by its ability to fit the experimental breakthrough
curves and, additionally, by whether  the experimentally determined model
parameters are realistic.
Experimental Apparatus
     The experimental apparatus used  for the  fixed-bed studies  is sketched
on Figure  10.  The major parts are:   a diffusion  cell assembly  to generate
an  influent gas  of  the  desired concentration,  flowmeters  to  regulate  the  flow
rate of  the influent gas  into the  charcoal bed, and  a gas  chromatograph
equipped with a  gas sampling valve to analyze both influent  and effluent  gas
streams.
Diffusion  Tube Assembly - The use  of  a  diffusion  tube  to  generate gas mixtures
containing organic  vapors  in  the  parts-per-million range  in  air has  been
described  by Altshuller and Cohen (2).   Our  diffusion  tube assembly,  shown
on Figure  11,  is patterned  after  their  design and consists of a long
 capillary  tube  connected to a liquid reservoir and open to a mixing chamber
at the top.   Liquid in the capillary tube  diffuses from the  liquid-gas
 interface  to  the upper end of the capillary  tube  at  a  rate given by the well
 known diffusion equation
                                    DPS       P                        rssv
                             rate = FT! ln P=?                       (55)
 Gas mixtures of desired concentrations could be prepared by  varying the gas
 flow rate across the upper end  of the diffusion tube,  or by  changing the
                                     30

-------
                          EXHAUSTf
         -*-
               GAS  SAMPLING
                VALVE
- o  -© -
  I
                     4 WAY
                     STOPCOCK

                 PRESSURES
                  GAUGE- ^
               —•X
GAS CRCjjflgTOGRAPH  ^

    RECORDER       J
              	vA
              O  O
              O  O
              O  O
              O  O

1
N




                    CONTROLLED  TEMPT.
                        WATER BATH
                          CONTROLLED
                          TEMPT
                          WATER BATH
                                            GAS
                                            MIXER
                                                         DRY AIR

                                             PRESSURE CONTROLLER

                                                        NITROGEN
                                                             GAS
   FKL 10
EXPERIMENTAL  ADSORPTION  SYSTEM

-------
ACETALDEHYDE
8 N2 MIXTURE
      1
  n
              ^/FRITTED GLASS
    MIXING  CHAMBER
    HYPODERMIC
    SYRINGE
    NEEDLE
    ACETALDEHYDE
    RESERVOIR
         UBBER STOPPER
         APILLARY
       FIG.  II
DIFFUSION  TUBE
                      32

-------
cross-sectional area of the diffusion tube, or the vapor pressure of the
diffusing component which depends only on temperature.
Chromatographic Analysis - In order to establish a calibration curve for
chromatographic analyses it was necessary to be able to generate gas mixtures
of known composition.  Since the air flow rate was easily controlled and
measured, it was necessary only to determine the rate of diffusion through
the diffusion tube.  For this calibration process a straight capillary tube
closed at the bottom end was used in the diffusion tube assembly.  The dif-
fusion rate was determined by measuring the change in liquid height inside
the capillary tube over a measured time interval.  The height change, measured
by means of a cathetometer, was converted to a volume by means of a previously
established calibration.  This calibration was established by partial
filling of the capillary with mercury followed by weighing on an analytical
balance and determination of the mercury column height with the cathetometer.
From the volume of liquid evaporated per unit time the diffusion rate was
determined from the known liquid density of the diffusing component.
     Chromatographic calibration curves were established by determining
chromatograms for gas mixtures of several different known concentrations.
The calibration curves for ethylmercaptan and acetaldehyde are shown in
the appendix.
Fixed-Bed Experimental Procedure
     A test was started by filling the diffusion tube and starting the gas
flows until a gas mixture of the desired composition could be maintained.
When constancy of the influent gas composition was assured the charcoal
holder was charged with a known weight of charcoal and the flow rate of
influent gas through the charcoal bed was regulated with the flowmeters.
Chromatographic analyses of influent and effluent streams were taken as the
test progressed.
Fixed-Bed Adsorption Studies with Acetaldehyde
     Preliminary tests showed that the acetaldehyde concentration of the gas
mixture produced by  the diffusion tube assembly could not be maintained con-
stant but continually decreased.  This was believed due to an oxidation process
occurring at the liquid-gas interface and nitrogen was substituted for air
as  the flowing gas.  With nitrogen flowing a constant diffusion rate was
observed, however, it was necessary  to mix the  small  stream  from the diffusion
tube assembly with air.  This was done in  the mixer shown on Figure  12.
                                     33

-------
                       ACETALDEHYDE
                       8 N2 MIXTURE
AIR
                           i
                     ^FRITTED GLASS
                         TO
                         CHARCOAL
                         BED
               FIG   12
GAS MIXER

-------
As shown in Figure 10 after leaving the mixer gas flowed through several
feet of tubing before reaching the flowmeters.  It is believed that thorough
mixing occurred.  The flow of nitrogen was much smaller than the air flow
so that when mixed with air only a small change in the oxygen to nitrogen
ratio occurred and the diluent gas could be regarded for all practical
purposes, as air.
     Breakthrough data were taken with an influent stream containing 28
parts per million of acetaldehyde in air.  Tests were made using various
influent flow rates, bed depths, bed heights, and charcoal particle sizes.
Pittsburgh PCB charcoal, derived from coconut shells, was used in all tests
and the adsorption temperature was maintained at 30°C for all tests.
Static Equilibrium Tests
     These tests were made so that the equilibrium capacity of the charcoal
could be determined and also because knowledge of the equilibrium relationship
is helpful in carrying out the modeling studies with the breakthrough data.
Flasks of 3 liters capacity were initially filled with known mixtures of
acetaldehyde and air.  Known weights of charcoal were then added and the
flasks were set aside at a controlled temperature of 30°C.  After two weeks
the gas in the flasks was analyzed.  To confirm the attainment of equilibrium
the gas was analyzed again several days later.  The two analyses were identical
which Indicated the existance of a state of equilibrium.  A material balance
calculation allowed the determination of the amount of acetaldehyde adsorbed
on the charcoal in equilibrium with the final acetaldehyde concentration in
the gas phase.  To check for error due to possible adsorption of acetaldehyde
onto the wall of the flask no charcoal was placed in one of the flasks.  The
acetaldehyde concentration within this flask was not observed to change
significantly indicating no adsorption on the flask wall.
FIXED-BED CATALYTIC STUDIES
     Before catalytic oxidation was discovered, or even expected, the
ethylinercaptan - air system was chosen for the study of fixed-bed adsorption.
The results of several such tests could only be explained in terms of a
chemical reaction involving oxygen and ethylmercaptan which is catalysed by
charcoal.  Because of these tests the fixed-bed catalytic study phase of
the project was initiated.
                                     35

-------
Experimental Apparatus
     The experimental apparatus was essentially the same as that used for
the fixed-bed adsorption studies and shown on Figure 10 with the exception
that it was not necessary to use a nitrogen flow through the diffusion tube
and therefore the gas mixer was not required.  Because considerably less
charcoal was needed for these studies than for the fixed-bed adsorption
studies it was necessary to mix the charcoal particles with glass beads in
order to obtain sufficient bed depth to insure good flow patterns through the
bed.  Usually, equal volumes of charcoal and glass beads were used to form
the bed.  Concentrations of ethylmercaptan in the influent and effluent streams
were again determined chromatographically.
Experimental Procedure
     The experimental procedure is very similar to that followed in the fixed-
bed adsorption studies; the bed is fed with an influent stream of constant
composition at a constant flow rate and the concentration of the effluent stream
is monitored.  A total of 13 fixed-bed tests were made during the study of
ethylmercaptan oxidation on Pittsburgh BPL charcoal catalyst.  The variables
studied ranged from  20 to 125 PPM  influent ethylmercaptan concentration, 3-40
cm/sec superficial velocity, and 3-14 gm  charcoal per cm  of ethylmercaptan
fed/min.  All except three of these tests were made with 14 mesh charcoal and
the remaining tests  were made with 8 mesh charcoal.
Identification of Reaction Products
     Several  attempts were made to collect and identify reaction products by
mass spectrographic  analysis.  Attempts using a liquid nitrogen trap on  the
effluent  stream were largely unsuccessful because oxygen and water, present
in  large  quantities  in the collected sample, did not  allow  the detection of
species which may have been present in much  smaller quantitites.  One  test
was made  with an acetone-dry ice  trap to  avoid condensing oxygen, but  the
quantity  of water present in the  sample again obscured  the  analysis.   Moderate
success was  achieved when a liquid nitrogen  trap was  connected  to  the  chromato-
graph  so  that the flow from the chromatograph  could be  diverted  through  the
trap when the "product" was scheduled to  appear.  Although  a fair mass spectro-
gram was  obtained,  the presence of air and water  limited  the degree of resolution.
 It  is  believed  that  the air entered  the system through  stopcocks on  the  trap
which  were not  of  the high vacuum variety and  the water had condensed  out  of
                                    36

-------
the nitrogen carrier gas used for the chromatograph.   Installation of high
vacuum stopcocks and drying of the carrier gas prior to its entering the
chromatograph should produce much better samples.
                                     37

-------
                               DISCUSSION
ACETALDEHYDE ADSORPTION STUDY
     By applying the previously discussed procedure breakthrough data were
obtained at 30°C for fixed-beds of Pittsburgh PCB charcoal fed with a con-
stant influent concentration of 28 PPM of acetaldehyde in air.  The effect
of bed length, superficial velocity, and charcoal particle size was studied.
The ranges of these variables were:  8 to 46 cm bed length, 10, 12, and 14
mesh charcoal particle size, and 8-40 cm/sec superficial velocity.  The
range of superficial velocities used coincides with the range customarily
used for design of commercial fixed-bed adsorption systems; Barnebey (3)
states that this later range is 35 to 75 ft/min (18 to 38 cm/sec).  The
experimental breakthrough curves are shown in the appendix and the salient
features of each test are tabulated in Table 1.
Charcoal Capacity
     The charcoal capacity was determined both from the static equilibrium
tests and from the breakthrough curves.
Static Equilibrium Tests - Nine equilibrium data points have been obtained
using the Static Equilibrium Method described previously.  These data, listed
in Table 2, allow the plotting of a smooth isotherm curve in the range of 0
to 35 ppm of acetaldehyde in air.  This plot is shown on Figure 13.
     Attempts were made to fit the isotherm data to Langmuir and Freundlich
isotherm equations.  If the Langmuir equation is followed, a plot of gas
concentration divided by equilibrium capacity versus gas concentration
should be a straight line.  Such a plot is shown on Figure 14 where it can
be seen that a reasonably good fit to a straight line occurs.  As a test
for  conformity of the data to the Freundlich isotherm equation, one plots
log  of equilibrium capacity versus log of gas concentration.   If a straight
line can be drawn on such a plot the data can be said to fit the Freundlich
isotherm equation.  Figure 15 shows a log-log plot of the isotherm data.
It could be said that the data fit a straight line reasonably well and
therefore also can be represented by the Freundlich isotherm equation.
Regeneration  studies - The regeneration of once-used Pittsburgh PCB
charcoal was  carried out in a vacuum oven at a temperature of  120°C and
                                     38

-------
u>
VO
                     TABLE  1.  RESULTS OF FIXED-BED TESTS WITH ACETALDEHYDE ADSORBED ON PITTSBURGH PCB  CHARCOAL


                                        Influent Concentration:  28 ppm of acetaldehyde in air

                                                   Temperature:  30°C
RUN No.
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
26
29
30
31
32
38
39
CHARCOAL SIZE
(mesh)*
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
14
10
12
BED WEIGHT
(GRAMS)
2.0
4.0
2.0
3.45
3.49
3.45
8.0
8.0
8.0
5.2
5.2
5.2
5.2
3.45
5.2
5.2
6.1
8.3
6.8
4.54
7.56
6.05
16.5
37.5
37.5
37.5
37.5
102
58
BED DIAMETER
(INCHES)
59/64
59/64
59/64
5/8
5/8
5/8
59/64
59/64
59/64
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
5/8
1
3/4
BED LENGTH
(INCHES)
0.5
1.0
0.5
2.0
2.0
2.0
2.0
2.0
2.0
3.0
3.0
3.0
3.0
2.0
3.0
3.0
4.0
5.375
4.5
3.0
5.0
4.0
8.0
18.0
18.0
18.0
18.0
18.0
18.0
INFLUENT
FLOW RATE
(cc/min)
1610
1300
1610
400
700
1000
1524
1524
879
700
400
1000
1000
1000
1425
1188
1425
1950
1950
1950
1425
1000
4700
1950
2970
4150
3560
4750
2750
SUPERFICIAL
VELOCITY
(cm/sec)
6.2
5.0
6.2
3.4
5.9
8.4
5.9
5.9
3.4
5.9
3.4
8.4
8.4
8.4
12.0
10.0
12.0
16.4
16.4
16.4
12.0
8.4
39.5
16.4
25.0
35.0
30.0
16.4
16.4
BREAK
TIME
(mln)
9
36
6
270
150
70
145
130
270
255
515
380
400
80
90
140
310
270
270
120
510
210
300
1100
738
375
468
884
1040
CHARCOAL
CAPACITY
(gm/gm)
0.00192
0.00192
0.00189
0.00407
0.00377
0.00303
0.00269
0.00365
0.00368
0.00381
0.00389
0.00793
0.00812
0.00320
0.00289
0.00405
0.00858
0.00751
0.00884
0.00759
0.00872
0.00585
0.00977
0.00722
0.00740
0.00650
0.00700
0.00580
0.00610
            *U.S.  Standard Selve Size

-------
TABLE 2.   RESULTS OF STATIC EQUILIBRIUM TESTS WITH ACETALDEHYDE
           AND PITTSBURGH PCB CHARCOAL AT 30°C
           Equilibrium Concentration       Charcoal Capacity
             (PPM of Acetaldehyde)              (gm/gm)

  !                  24.0                       0.00718
  2                  20.0                       0.00592
  3                  21.5                       0.00653
  4                  15.5                       0.00582
  5                  35.0                       0.00869
  6                  28.5                       0.00839
  7                   5.5                       0.00406
  8                   3.0                       0.00247
  9                   1.5                       0.00125

-------
   O.OI
   0.005
o
o
                           10
15
 20        25        3O         35
ACETALDEHYDE CONCENTRATION, ppm
40
            FIG.   13     STATIC  EQUILIBRIUM  DATA  FOR ACETALDEHYDE ON
                        PITTSBURGH  PCB  CHARCOAL  AT  30 °C

-------
      8 r
<0
 O
                                                C x 10
           FIG.  14  . EQUILIBRIUM DATA  FOR  ACETALDEHYDE ON PITTSBURGH

                     PCB CHARCOAL IN LINEAR  FORM

-------
 E
 a
V.

 E   -0

3  102
u
               FIGURE  15     TEST OF  THE  FREUNDLICH

                            ISOTHERM  EQUATION
u
3
u
Of
u
   10'
                              10
                 PPM  OF  ACETALDEHYDE
100

-------
a vacuum of 20 inches Hg.  Charcoal regenerated under these conditions for
25 hours was used for a fixed-bed test where a significantly lower capacity
was observed.  The charcoal was then subjected to an additional 100 hours of
regeneration at the above specified conditions and again used for a fixed-bed
test.  A sizeable increase in capacity was observed although it was somewhat
lower than that found from fixed-bed tests with virgin charcoal.  The
charcoal was then subjected to an additional 245 hours of regeneration.
This charcoal showed a fixed-bed capacity in the range of the virgin char-
coal and was judged to be completely regenerated.
Fixed-bed  tests - Examination of Table 1 reveals that although the influent
concentration is the same for all tests  (28 PPM) the charcoal capacity is
not  constant.  This was  an unexpected development because we had found no
mention in the literature of any study where adsorbent capacities were
reported to  be inconsistent.  This discrepancy was disconcerting and a
major part of our efforts were devoted toward resolving it.  We felt  that
if we were unable to predict the capacity of a charcoal bed  there was very
little  chance of developing a procedure  for design of charcoal beds.
     The inconsistency  in charcoal capacity is not believed  due to experi-
mental  technique because it was observed that  there  is little  scatter  in
 the  experimental data  points and a smooth breakthrough curve can be drawn
 in all  cases.  Further,  several  tests were  duplicated and  these duplicate
breakthrough curves  were found to  superimpose  quite  well.
     Two different diameter beds were  used  and it was found  that  for  identi-
 cal  superficial  velocities  the breakthrough curves  and hence the  charcoal
 capacities were  identical.  The  fact that the  breakthrough behavior was  not
 found  to  depend  upon bed diameter  is not surprising because  for all bed
 diameters  investigated the  ratio of  bed  to  particle diameter was  greater
 than 8, the generally accepted minimum value  for packed  beds.
      Most  of the tests were made to  elucidate the effect of bed length and
 superficial velocity upon charcoal capacity.   Although  no distinct relation-
 ship among these variables  was  found,  we can state that  for a  constant
 superficial velocity the charcoal  capacity increases with increasing  bed
 length, but approaches a value of  about  0.0088 gm/gm.  as an upper limit.
 It is significant that  this upper limit  is very close to the estimated
 equilibrium capacity of charcoal as read from Figure 13.
                                     44

-------
     The dependence of capacity on bed length is believed due to two factors;
flow patterns within the charcoal bed and the very slow rate of mass transfer
between the gas phase and the solid phase when solid-phase saturation is
approached.  Although the existence of the first effect can be experimentally
demonstrated, it is believed that the second effect is the major factor.
     In an attempt to determine the effect of bed depth on flow patterns
pressure drops across beds of Pittsburgh PCB charcoal (14 - 16 mesh) of
various depths were determined at several air flow rates.  The bed was
formed inside a 16 mm O.D. glass tube fitted with a coarse fritted glass
disc in the bottom for distribution of the incoming air.  This is the same
size of tube and type of fritted glass disc as was used in our fixed-bed
studies.  Pressure changes across the fritted glass disc and charcoal bed
were determined with a pressure transducer.  The pressure transducer was not
standarized because only relative readings were deemed necessary.
     The results are plotted on Figure 16 as relative pressure units versus
bed depth at each superficial velocity.  One would expect these data to plot
a single straight line for each velocity if flow patterns were independent
of bed length.  This seems to be the  case for U = 17 cm/sec and U = 21.25
cm/sec, however, at lower velocities  this expected linearity is not found
until a bed depth of  3 or 4 inches  is reached.  This would indicate that
flow patterns within  the bed do change as bed depth changes.
     Changes in charcoal capacity could result  from changes  in flow patterns
when some  of the channels and void  spaces in  the bed did not receive an
active flow of gas and became  stagnant.  The  charcoal  in  these stagnant
sections of  the bed would adsorb at an extremely slow  rate because  the
adsorbate  would reach the adsorbent surface only after diffusing  through
stagnant air in these channels.  The time required for this  diffusion  to  the
adsorbent  surface  could  be large compared with  the time required  to obtain
the  breakthrough data and thus  it would  appear  that the capacity  of  the
charcoal  in  the bed was  less  than  the equilibrium  capacity.
     When  the  adsorption rate is  controlled  by  a  solid phase resistance
 the  rate  is  known  to  slow considerably as  the solid phase approaches  satur-
ation.   This type  of  resistance is  characterized  by a breakthrough  curve
which  rises  rapidly at  low values  of c/c  and flattens out as  higher values
of  c/c  are  approached.   An  examination  of  Figures A-l through A-13 reveals
                                     45

-------
    u= 21.25 cm/sec
  u=l70cm/sec
                                 Fig.16
                                Pressure Drop Through
                                Beds of Charcoal
246
   Bed Length (in)   ,

-------
that all of our breakthrough curves exhibit this characteristic shape and
therefore postulation of the solid-phase resistance controlling the rate of
adsorption seems quite reasonable.
     If the charcoal bed is short then the exhaustion point occurs after
only a small quantity of fluid has passed through the bed.  Although the
charcoal in the bed still has the potential to adsorb, the rate is quite
low and gas flowing through the bed has a short contact time with the solid.
With this short contact time and the low rate of adsorption the gas composi-
tion changes very little and the bed gives the appearance of exhaustion.  If
the contact time is lengthened by lengthening the packed bed a detectable
gas composition change is possible even with low rate of adsorption.  Theo-
retically, if the bed is long enough the capacity determined from the break-
through curve should be the equilibrium capacity of the adsorbent.  From
the results of our static equilibrium studies the capacity of a charcoal bed
in equilibrium with an acetaldehyde-air mixture containing 28 ppm of
acetaldehyde is about 0.0084 gm/gm.  This agrees quite well with the maximum
value of the charcoal capacity calculated from the breakthrough curves of
about 0.0088 gm/gm.
Modeling Study
     The objective of the modeling study is to establish the rate mechanism
and to determine the dependence of the parameters in the rate expression upon
the system variables so that the design of adsorption systems for odor removal
can be accomplished.  By fitting a selected model breakthrough equation to
the experimental breakthrough curve, the evaluation of the model parameters
is possible.  The validity of a given model is then determined by its ability
to fit the experimental breakthrough curve and also by whether the experimentally
determined model parameters are realistic.  With design as our prime objective,
we want fairly simple models whose parameters have physical significance and
can be readily estimated.  The four model equations developed in the Theory
Section of this report were applied to the experimental breakthrough data.
Because of the inconsistency of the charcoal capacity found for beds of depths
less than three inches, only the tests mady with bed depths three inches or
greater were used for the modeling study.
     All of the four model equations contain as parameters the number of
transfer units, the separation factor, and the distribution ratio.  The
                                    47

-------
number of transfer units is defined by equations 20, 26, 33, and 49, the
separation factor is defined by either equation 15 or 16 and the distribu-
tion ratio, D, appears in equation 18 which defines the through-put para-
meter Z and is defined as
                                    q* W
                               D - cTTT                            (56>
                                    o
The separation factor and the distribution ratio are seen to be expressable
in terms of the equilibrium data and it is normally expected that they can
be evaluated if the equilibrium data are known.  In our first attempt to
fit the model equations to the experimental breakthrough data we evaluated
these parameters from our equilibrium data shown on Figure 13.  We were
unable to obtain a reasonable fit and therefore decided to use the separa-
tion factor as evaluated from our equilibrium data (R = 0.28), but regard
the distribution ratio as a parameter to be obtained from the curve fitting
process.  When used in this empirical manner, D will be denoted by D  and
called the effective distribution ratio.
     All model equations were fitted to the experimental breakthrough data
with a least-squares procedure.  Because of the extreme "tailing off"
exhibited by most of our breakthrough curves, the upper portion of the
experimental breakthrough curve was not used in the curve fitting process.
Figures A-l through A-13 show a comparison of the experimental breakthrough
curve and the various model equations which were fitted to the breakthrough
data.  The parameters obtained from fitting each of the four model break-
through equations are listed in Tables 3 through 6 and are plotted versus
the residence time, ve/F, in Figures 17 through 24.
     Inspection of Figures A-l through A-13 reveals that of the four model
equations the particle-phase diffusion (or solid diffusion) model fits the
breakthrough curves over the widest range of effluent concentration.  All
model equations fit rather well in the region from about 5 to 17 PPM,
however for most tests the solid diffusion model fits over the wider range
0 to 17 PPM.
Particle-phase Diffusion Model - Figure 17 shows that the effective distri-
bution ratio can be considered constant.  Even though this average value
is only a fraction of the value which one calculates from the equilibrium
                                    48

-------
TABLE 3.   PARAMETERS OBTAINED FROM MODEL EQUATION FOR
           PARTICLE-PHASE DIFFUSION MODEL
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
ve/F
sec
0.677
0.595
0.592
0.515
0.303
0.747
0.556
1.471
0.966
0.691
0.745
1.239
1.401
Effective Dist-
ribution Ratio
D
ep
62,386
57,990
59,179
54,441
59,020
68,341
54,002
53,159
58,127
47,780
51,179
52,872
54,577
N
P
5.00
4.00
3.60
4.80
3.00
4.60
6.30
15.00
10.00
5.25
5.98
10.00
11.00
                          49

-------
TABLE 4.   PARAMETERS OBTAINED FROM MODEL EQUATION FOR
           PORE DIFFUSION MODEL
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
ve/F
sec
0.677
0.595
0.592
0.515
0.303
0.747
0.556
1.471
0.966
0.691
0.745
1.239
1.401
Effective Dist-
ribution Ratio
D
epore
60,746
55,543
54,146
55,441
53,650
63,739
54,009
52,624
56,130
44,064
48,199
50,385
54,321
N
pore
3.081
2.373
2.384
2.355
1.985
3.349
3.188
10.990
8.827
10.040
8.768
10.769
6.782
                          50

-------
TABLE 5.  PARAMETERS OBTAINED FROM EXTERNAL DIFFUSION
          MODEL EQUATION
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
ve/F
sec
0.677
0.595
0.592
0.515
0.303
0.747
0.556
1.471
0.966
0.691
0.745
1.239
1.401
Effective Dist-
ribution Ratio
Def
53,000
47,457
46,644
45,252
44,281
57,431
47,915
50,954
54,125
42,703
46,454
48,960
51,341
Nf
5.026
4.164
3.363
4.081
2.685
4.807
7.466
26.261
13.320
11.806
13.846
11.331
16.796
                          51

-------
TABLE 6.   PARAMETERS OBTAINED FROM REACTION KINETICS
           MODEL EQUATION
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
ve/F
sec
0.677
0.595
0.592
0.515
0.303
0.747
0.556
1.471
0.966
0.691
0.745
1.239
1.401
Effective Dist-
ribution Ratio
DeR
57,607
52,437
52,703
50,097
51,485
62,651
50,719
51,801
55,901
44,284
47,919
50,847
52,677
NR
7.85
6.28
5.35
7.35
4.59
8.33
11.11
32.07
18.52
17.71
21.33
20.83
21.35
                         52

-------
   80,000
O  60,000
   40,000
CO
a
u
>  20,000
u
u
 _
fc
                               0.5
         1.0

MEAN RESIDENCE TIME (ve/F.sec)
1.5
                 FIG.  17   .  RELATION BETWEEN  MEAN RESIDENCE TIME  AND
                            EFFECTIVE DISTRIBUTION RATIO FOR   PARTICLE-PHASE
                            DIFFUSION  MODEL .  AVERAGE VALUE OF Dep= 57,865

-------
SU£)UU
§ eopoo
o"
1
P 4OPOO
CD
oc
5
UJ
g zopoo
CJ
0
(
\
-
-



—
, i 1 i 1 I i i i i i i 1 J 	 1
3 O.5 1.0 1-5
                                MEAN RESIDENCE  TIME (v«/F, sec)

FIG.  18  . RELATION BETWEEN MEAN RESIDENCE TIME AND EFFECTIVE
          DISTRIBUTION  RATIO  FOR PORE DIFFUSION MODEL
          AVERAGE VALUE OF  Depore= 53.2IO

-------
eopoo
o
0
P
* 40,000
O
P
m
oe
CO
5 2OtOOO
ui
o
£
ijj _
<
•
"
* * •
~

-

i i i i i i i i i i i i i i i
) O.5 I.O 1.5
                                 MEAN RESIDENCE TIME (ve/F, sec)

FIG.  19   . RELALATION BETWEEN MEAN RESIDENCE TIME AND  EFFECTIVE
           DISTRIBUTION  RATIO FOR EXTERNAL DIFFUSION  MODEL
           AVERAGE VALUE OF  Def = 47,607

-------
   eopooi-
   60,000
O
<
O 4O.OOO
O
I
UJ
  2O,OOO
                               05
          1.0
MEAN RESIDENCE TIME ( ve/F, sec )
                                                                                1.5
             FIG.  20   .  RELATION  BETWEEN  MEAN RESIDENCE TIME  AND  EFFECTIVE
                         DISTRIBUTION RATiO  FOR  REACTION  KINETICS MODEL
                         AVERAGE VALUE OF  DCr=5l,52O

-------
  16
CO
  14
gj 12
UL
CO


K 10
CO
to  _
<  8
§4
                          0.5
               1.0


MEAN RESIDENCE TIME  ( ve/F , see )
1.5
          FIG.  21   .  NUMBER OF MASS TRANSFER UNITS VERSUS RESIDENCE


                     TIME  FOR PARTICLE-PHASE DIFFUSION MODEL

-------
       N
        pore
             12
             to
             8
             6
00
                                 0.5
1.0
1.5
                                                         ve/F, sec.
                  FIG.  22  ,  NUMBER OF MASS TRANSFER  UNITS  VERSUS RESIDENCE
                             TIME FOR  PORE DIFFUSION  MODEL

-------
28
24
20
16
12
 8
                                     1.0
                                         ve/F, sec
1.5
  FIG.  23   .NUMBER OF MASS TRANSFER UNITS VERSUS MEAN
             RESIDENCE  TIME  FOR EXTERNAL DIFFUSION MODEL
                          59

-------
35 r
30
25
20
15
10
      i    i
                 -J	U
                                                           1.5
0                 0.5                  1.0
                                            ve/F, sec

FIG.  24   . NUMBER OF  MASS TRANSFER  UNITS  VERSUS RESIDENCE
           TIME FOR REACTION  KINETICS MODEL
                        60

-------
data, the constancy of the parameter is expected and in this respect it
can be said that the parameter behaves realistically.
     The particle-phase mass transfer coefficient, k , is not expected to
depend upon superficial velocity or the residence time, ve/F, and therefore
equation 26 predicts that N  should be a linear function of ve/F.  The plot
of this parameter, Figure 21, shows a reasonably good linear relationship,
however, the solid curve appears to fit the data somewhat better.  In a
qualitative sense this parameter also behaves realistically.
     Using smoothed values of the two parameters as obtained from the solid
curves on Figures 17 and 21 the breakthrough times (time when C = 0.05 C )
                                                                        o
were back-calculated breakthrough times agreed with the experimental values
to within 10 percent; N  values obtained from the smooth curve on Figure 21
gave better results than values obtained from the linear relationship.  A
comparison of these experimental and calculated results is shown on Figure 25
and details of the calculation procedure are given in the appendix.
Pore Diffusion Model - As shown by Figure 18, the effective distribution
ratio is constant as expected.  Although the effective distribution coeffi-
cient is considerably smaller than the value calculated from equilibrium
data, it at least shows qualitatively realistic behavior.
     Because k     is expected to be independent of ve/F. a plot of N
              pore      r                ^            iff        pore
versus ve/F should be linear.  Figure 23 shows that while there is consider-
able scattering to the data a linear relationship is not unreasonable.
However, it was found that parameters obtained from the solid curve gave
better results for the back calculation of breakthrough times.  In addition
to this qualitative evaluation it is possible to estimate the pore diffusi-
vity and hence calculate values of N     corresponding to the conditions
                                    pore             °
of each breakthrough curve.  Wheeler (4) has derived the following equation
for the calculation of pore diffusivity
D
        V
                 pore
                                              8Rg
                                                M

                                                                     (57)
From the values of the internal void fraction, x , and average pore radius,
r , supplied by the Calgon Corporation (5) for Pittsburgh PCB charcoal the
pore diffusivity, D   e> was calculated.   This value was then used to
calculate N     via equations 47 and 49.   Values of N     so calculated
           pore                                      pore
                                     61

-------
o
z

Id
   eopoor
   70,000 -
   60,000 -
   50,000 -
   40000 -
 O
 u
 X
   30POO.
   20000 -
   10,000 -
               10,000   20,000   30POO   40,000   50,000   60000   70,000


                               W05, see  IEXPERIMENTAU
          FIG.  25  .  COMPARISON OF PARTICLE-PHASE MODELING BREAK-

                     TIMES WITH  EXPERIMENTAL  VALUES

-------
are compared with the N     values obtained from the modeling study in
Table 7 where it is seen that the agreement is remarkably good.
     Results of the back-calculation of breakthrough times using smoothed
parameters from Figures 18 and 22 are shown on Figure 26.  Again the back-
calculated values agree with experimental values to within 10 percent.
Fluid-phase Diffusion Model - Figure 19 shows that the effective distribution
ratio is again constant and hence qualitatively realistic.
     Because the fluid-phase mass transfer coefficient will be a function of
superficial velocity and hence not constant, no simple dependence of Nf
upon ve/F is expected.  However, it is possible to estimate the mass
transfer coefficient and hence calculate N. for the conditions of each
breakthrough curve.  The correlation recommended by Wilke and Hougen (6)
was used to estimate kfa and Nf was calculated via equation 20.  The com-
parison of these estimated Nf values with those obtained from the model
study is shown in Table 8 where it is seen that only fair agreement is
obtained.
     The back-calculation procedure was carried out with smoothed parameters
from Figures 19 and 23 again obtaining agreement within ten percent with the
experimental values.  A comparison is shown on Figure 27.
Reaction Kinetics Model - As shown on Figure 20 the effective distribution
ratio shows the same behavior as was found with the other model equations.
     It has been shown that the number of reaction units, ND, can be inter-
                                                           K.
preted as being dependent upon both solid-phase and fluid-phase resistances.
For this reason it is not expected that N  will show a simple dependence on
                                         K
superficial velocity or ve/F.  That this is indeed the case can be seen
from Figure 24.  Thus, there appears to be no criterion by which Nn can be
                                                                  K
judged to be realistic except to say that N  shows the same range of values
                                           K
as does N  and N...  This is to be expected if N  is visualized as arising
         p      r                              K
from the addition of resistances in series.
FIXED-BED CATALYTIC STUDY
     The results of fixed-bed tests made with Pittsburgh BPL and PCB charcoal
and ethylmercaptan are tabulated in Table 9 and a plot of conversion versus
W/F for Pittsburgh BPL charcoal is shown on Figure 28.  All data were taken
at atmospheric pressure and a temperature of 30°C.
                                    63

-------
TABLE 7.   COMPARISON OF CALCULATED N     VALUES WITH MODELING
           VALUES                    P°re
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
d
mm
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
2.00
1.68
Pe
11.61
16.54
22.63
22.63
22.63
16.54
54.54
22.63
34.47
48.20
41.36
32.80
27.55
Calculated
N
pore
3.59
2.95
2.94
2.41
1.61
3.66
2.42
13.29
8.73
6.24
7.27
7.20
10.04
Modeling
N
pore
3.00
2.68
2.68
2.40
2.00
3.26
2.55
10.95
4.85
3.04
3.25
7.90
10.00

-------
TABLE 8.   RESULTS OF CALCULATED N  VALUES COMPARED WITH THE
           MODELING Nf
Run No.
APC-14
APC-18
APC-19
APC-20
APC-21
APC-22
APC-26
APC-29
APC-30
APC-31
APC-32
APC-38
APC-39
D
nun
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
1.41
2.00
1.68
P
e
11.61
16.54
22.63
22.63
22.63
16.54
54.54
22.63
34.47
48.20
41.36
32.80
27.55
Calculated
Nf
6.35
6.22
7.22
5.92
3.95
7.71
9.17
32.65
26.34
22.20
24.00
21.22
27.15
Modeling
Nf
4.6
3.9
3.9
3.4
2.6
5.6
3.7
26
9.9
4.9
5.6
18
23.3
                             65

-------
TABLE 9.   RESULTS OF FIXED-BED TESTS WITH ETHYLMERCAPTAH AND PITTSBURGH BPL CHARCOAL
Run No.
M-l
M-2
M-3
M-4
H-5
M-6
M-7
M-8
M-9
M-10
M-ll
M-12
M-13
M-14
4
Charcoal
Type
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
BPL
PCB
Charcoal
Size
(mesh)
14
14
14
14
14
14
14
14
14
14
14
8
8
8
-
Charcoal
Weight
(gm)
1.0
1.0
1.0
0.5
1.0
0.75
1.0
0.5
1.0
1.0
1.0
2.0
1.0
1.0
2.0
Influent
Flow Rate
(cc/min)
720
4700
4700
4700
4700
2800
2800
2800
2800
2800
1250
4700
4700
4000
710
Influent
Concentration
(PPM)
20
20
20
20
20
60
60
60
25
45
125
20
20
50
27
Conversion
%
100
77.5
82.5
30.0
80.0
70.0
61.6
41.7
83.0
70.0
69.0
86.5
78.0
59.0
100
Superficial
Velocity
(cm/sec)
2.7
39.5
39.5
39.5
26.6
15.8
15.8
15.8
15.8
15.4
7.05
20.6
20.6
17.5
__
W/F
, Km charcoal *
cc of reactant/mln
69
10.6
10.6
5.3
10.6
4.46
5.95
2.97
14.3
7.93
6.40
21.3
10.6
5.00
104

-------
    80,000 r
    70,000 -
    60,000-
z   50,000
ui
Q
O
    40,000-
    30,000 -
    2QPOO -
    10,000 -
               10,000   20,000   30,000   40,000  50000   60,000  70,000


                                    t*=0.05  , sec (EXPERIMENTAL)




           RG.  26   .  COMPARISON OF  PORE  DIFFUSION  MODELING

                       BREAK-TIME  WITH  EXPERIMENTAL  VALUES
                             67

-------
    70,000 -
    60,000 -
^  50POO

o
Ul
o
   40 POO -
g  30POO
   20,000 -
    10,000
         >      10,000   20,000   30,000   40pOO   50,000   60pOO   70,000


                                        ^=0.05, MC (EXPERIMENTAL)




          FIG.  27  .  COPARISON  OF EXTERNAL  DIFFUSION  MODELING


                     BREAK-TIMES WITH EXPERIMENTAL  VALUES
                                68

-------
   70,000
   60,000
   50,000
o
••
u  40,000
i
 •. 30,000
o
d
 n
   20,000
    topoo
         0      10,000   20,000   30,000   40,000  50,000   60,000   70.OOO
                                          **=0.05,sec (EXPERIMENTAL)

         FIG.  2 8  . COMPARISON OF REACTION  KINETICS MODEL  BREAK-
                    TIME DATA   WITH  EXPERIMENTAL VALUES
                                 69

-------
     The data from a typical fixed-bed test are plotted on Figure 29.  As
shown on the figure, the effluent concentration of ethylmercaptan quickly
attains its steady state value and after considerable time a new peak
appears in the chromatographic analysis of the effluent stream.  This new
peak is believed to represent one or more reaction products which are
apparently adsorbed onto the charcoal and begin to desorb as the charcoal
becomes saturated.  When saturation is approached, product is observed in
the effluent stream with a concentration history closely resembling a con-
ventional breakthrough curve.
     Test M-3 was allowed to continue for seven days in order to observe
whether the charcoal looses its catalytic efficiency with time; no indica-
tion of this was found.  For the duration of the test the 1 gram of charcoal
had removed 2.15 gram of ethylmercaptan.  An extrapolation of published
equilibrium data for ethylmercaptan on a similar charcoal in the absence of
air (7) indicates that when saturated with respect to our influent stream
(20 PPM) each gram of charcoal should adsorb about 0.03 grams of ethylmercaptan.
Relative to its equilibrium adsorption capacity it is seen that the charcoal
has catalyzed the oxidation of a large quantity of ethylmercaptan without
loss of catalytic activity.
Mechanism Study
     Before the reaction mechanism can be discerned from the fixed-bed data
it is necessary to ensure that the reaction is not controlled by either external
or internal diffusion.
External Diffusion Test - Tests M-2, M-3, and M-5 were all made with a bed
containing 1 gram of charcoal and fed at the rate of 4700 cc/min of an
influent stream containing 20 PPM of ethylmercaptan in air.  Tests M-2 and
M-3 were made with a bed geometry which resulted in a superficial velocity
of 39.5 cm/sec.  These tests showed conversions of 82.5% and 77.5% which
indicates the degree of reproducibility obtainable.  The test M-5 was made
with a bed geometry which resulted in a superficial velocity of 26.6 cm/sec
and showed a conversion of 80.0%.  Because the conversion found for test
M-5 is essentially the average of that found for the other two tests it can
be said that a change in superficial velocity did not significantly affect
the conversion.  Thus external diffusion is not a rate-controlling resistance
in this range of superficial velocity.
                                    70

-------
                                                                                                    100
                    1000
                                2000

                          TIME (MINUTES)
9000
                                                                                             4OOO
FIGURE  29.
TYPICAL  EFFLUENT  HISTORY  CURVES  FOR THE   CATALYTIC  OXIDATION OF  ETHYLMERCAPTAN.
BY  CHARCOAL.

-------
Internal Diffusion Test - As can be seen from Figure 30 all data points
fit a single curve when conversion is plotted versus W/F.  Included on
this plot are three data points representing tests made with 8 mesh
charcoal thus indicating that the rate is unaffected by particle size.
If internal diffusion were the rate-controlling step, the rate would be
expected to depend upon the inverse square of the particle size and
therefore a noticeable decrease in conversion should have been observed
for the larger particle size.  As this is not the case it appears that the
rate of reaction is not limited by internal diffusion.
Test for Reaction Mechanism - Table 9 shows that essentially total conver-
sion was found for test M-l; with this information it is not necessary to
consider reaction mechanisms involving a reverse reaction term.  The
integrated reaction equation approach described in the Theory section of
this report was used to analyze the data.  Several simple reaction mechanisms
were tested, but a one-half order reaction was found to best fit the data.
For this mechanism the integrated rate equation is given by equation 53
which indicates that a straight line should result from a plot of
 1/2         1/2
C   [l-(l-x.)   ] versus W/V.  It is to be further noted that on such
a plot the data for all values of C  should lie on a single straight line
passing through the origin.  The data are plotted in this fashion on
Figure 31 where it is seen that while there is some scattering, with the
exception of the 125 PPM data point which lies considerably below the line,
these conditions are satisfied.  This low conversion found for the 125 PPM
point might be due to the extremely low superficial velocity used for this
test which could result in the reaction being controlled by external
diffusion.
     Hougen and Watson (8) have shown that a catalytic reaction one-half
order in a reactant can be explained in terms of a mechanism whereby that
reactant disassociates on the catalyst before participating in the reaction.
Further, the effluent composition history curves for individual tests,
such as shown on Figure 29, suggest that the reaction rate is not limited
by the number of active sites because the rate of decomposition of reactant
is unaffected by the buildup of product on the charcoal.  This is probably
the reason that the rate can be represented with a simple one-half order
expression rather than requiring the more complex forms suggested by the
active-site theory.

                                    72

-------
  100
I
o
                                                 cm9 Of REACTANT/mm.
            FIGURE 30.  A PLOT 'OF  CONVERSION VERSUS W/F FOR FIXED-BEDS
                       OF PITTSBURGH BPL CHARCOAL FED WITH AW INFLUENT
                       STREAM OF  ETHYLMERCAPTAN  IN AIR.

-------
 I
 a.
JT—
13	
  x
  i
 —  2 -
m
d e
 U
                                                                         O  20-25  PPM
                                                                            45-60 PPM
                                                                             J25 PPM
                                                   cc.
              FIG  31.   TEST  OF 0.5  ORDER  REACTION

-------
     The above analysis applies to Pittsburgh BPL charcoal catalyst; the
single fixed-bed test made with Pittsburgh PCB charcoal is insufficient to
allow a similar evaluation of its catalytic effectiveness.  It can only be
said that this type of charcoal also possesses catalytic activity.
Reaction.Products Identification
     The mass spectrogram obtained from the sample which was collected by
the trap on the chromatograph showed the presence of diethyldisulfide and some
other peaks which could not be identified.  Because of uncertainty associated
with the other mass spectrographic analyses, diethyldisulfide appears to be
the only reaction product we are able to identify.  Since disulfides are
known to be products of the air oxidation of mercaptans (9,10) the presence
of diethyldisulfide in the reaction products is entirely reasonable.
Neutron Activation Analyses of Charcoal
     An analysis of trace elements present in charcoal using neutron activation
analysis was undertaken with the hope that this information might help explain
the catalytic activity.  Three types of charcoal were analyzed:  Pittsburgh
BPL, Pittsburgh PCB, and Barnebey-Cheney AC.  The first is a coal-based
charcoal and the others are derived from coconut shells.  The results, listed
in Table 10, were determined by subjecting the charcoal samples to differing
degrees of neutron bombardment and determining their spectra after various
decay times.  It was originally intended to determine the concentration of
each element, however before this could be accomplished it became necessary
to change detectors and thereby the calibration was lost.
                                     75

-------
TABLE 10.  ELEMENTS PRESENT IN CHARCOAL SAMPLES AS DETERMINED
           BY NEUTRON ACTIVATION ANALYSIS
  Element
Pittsburgh     Pittsburgh      Barnebey-Cheney
    BPL            PCS         	AC	
Sodium
Manganese
Vanadium
Potassium
Chlorine
Aluminum
Cobalt
Iron
Scandium
Cesium
Rubidium
Barium
Gold
Lanthanum
970 ppm
288 ppm
119 ppm
X
X
2541 ppm
X
X
X
X
X
X
X
X
740 ppm
224 ppm
0
X
X
X
0
X
0
X
0
0
0
0
313 ppm
140 ppm
36 ppm
X
X
X
0
0
0
	
	
	
	
	
         X denotes the presence of an element

         0 denotes the absence of an element

       	 indicates that a test was not made

-------
                          APPLICATION TO DESIGN
ACETALDEHYDE REMOVAL
     The operation of a fixed-bed adsorber is cyclic.  First, a bed of
freshly regenerated adsorbent is contacted with the influent stream con-
taining the adsorbate and operated until the adsorbate concentration in
the effluent stream reaches some predetermined level, normally about 5%
of the influent value.  The bed is then regenerated in preparation for
another adsorption cycle.  Thus the design of this type of system requires
that both the adsorption and desorption behavior be known and understood.
     The ultimate design objective would be a priori approach where the
mathematical description of the system could be formulated then transformed
into useable form from a knowledge of adsorbent properties such as particle
size, average pore size and bed porosity, fluid properties such as viscosity
and diffusivity, and system variables such as temperature, pressure, and
fluid flow rate.  In addition to the often intractable mathematics, problems
arise because some of the phenomena are often not completely understood
nor is the basic physical data available to implement the computations
for those phenomena which are understood.  Failing in this a priori approach
the next best strategy appears to be the modeling approach.  Here, small
scale experiments with the proposed adsorbate-adsorbent systems are required
before sufficient understanding and mathematical description for design
purposes are possible.
     The original intent of this study was to conduct enough small scale
experiments for the acetaldehyde-Pittsburgh PCB charcoal system so that
a realistic model could be selected and the model parameters (e.g. distri-
bution ratio, separation factor, and number of transfer units) evaluated
and correlated against the pertinent system variables.  From our experiments
it appears that the solid-phase diffusion model and the pore diffusion model
are realistic and could safely be used to design a fixed-bed adsorber
operating at 30°C with an influent stream containing 28 PPM of acetaldehyde.
While the separation factor is known and the number of transfer units can
be estimated, design for other influent concentrations would involve
considerable uncertainty about the value of the effective distribution
ratio.
                                     77

-------
     The design of an adsorber fed with an influent stream containing 28 PPM
of acetaldehyde and operated at 30°C can be accomplished by using the model
breakthrough equations with the smoothed parameters or if the conditions
are equivalent, by directly employing the experimental data.  This is
possible because the model equations are written in dimensionless form.
Thus, for  the same charcoal particle size and superficial velocity the
break time obtained from a small, laboratory scale bed would apply for a
bed of much larger dimensions providing that the ratio v/F is the same
for both.
     The design of an adsorber fed with an influent stream containing between
35 and 1 PPM of acetaldehyde and operated at 30°C can be accomplished via
the solid-phase diffusion model equation, equation 25, if the effective dis-
tribution ratio can be estimated for influent concentrations other than 28
PPM.  The distribution ratio is defined by equation 56, however, recall that
it was found necessary to use an effective distribution ratio which was
determined by fitting the breakthrough data to a model equation.  For the
solid-phase diffusion model a constant value of 57,865 was found for the
influent concentration of 28 PPM.  Since w/ve is expected to be constant
for a given charcoal, equation 56 suggests that an effective distribution
ratio for other influent concentrations would be
                                       a*
                                       1-     00
                        °e = 57'865 tcU)^  ^
                                               o
where q* and C  represent a point on the equilibrium curve,  Figure 13,
corresponding to the desired influent concentration, C .
                                                      o
     The typical adsorption system would probably consist of a set of three
fixed-beds:  one operating, one regenerating, and one cooling.  The required
amounts of charcoal corresponding to various removal percentages and influent
concentrations are shown in Table 12 for the case where each bed spends one
hour on stream and two hours regenerating and cooling.  An average superficial
velocity of 50 ft/min (reference 3)  was used for these calculations.   Until
the regeneration process is better understood,  more detailed calculations
do not seem to be justified.   However,  the data of Table 12  indcate that
fixed-bed adsorption appears to be a feasible means of removing acetaldehyde
vapor from low concentrations in air.
                                      78

-------
ETHYLMERCAPTAN REMOVAL
     The rate of ethylmercaptan oxidation on charcoal was not found to depend
on superficial velocity or on charcoal particle size and therefore a reactor
can be designed from equation 53, the integrated rate equation.   The value
of k determined from the slope of the straight line on Figure 31 is 0.117
     1/9   1/2    -1            -1
gmole    cc    rain   gm charcoal  .   With this k we can determine the mass
of catalyst required to achieve any conversion at 30°C and for any influent
concentration in the range 20 to 125 PPM.  In Table 11 are listed the pounds
of charcoal required to remove various percentages of ethylmercaptan from
100 cfm of air containing various influent concentrations.
                                       79

-------
TABLE 11.  AMOUNT OF CHARCOAL REQUIRED TO REMOVE
           VARIOUS PERCENTAGES OF ETHYLMERCAPTAN
           FROM VARIOUS CONCENTRATIONS IN AIR AT
           30°C.
  Influent
Concentration      Pounds of Charcoal per 100 cfm Air
   (PPM)                90%       95%
     25                 2.32      2.64      3.06

     50                 3.26      3.72      4.30

     75                 4.01      4.56      5.29

    100                 4.61      5.25      6.09

    125                 5.15      5.86      6.80
                            80

-------
TABLE 12.  AMOUNT OF CHARCOAL REQUIRED TO REMOVE VARIOUS
           PERCENTAGES OF ACETALDEHYDE FROM VARIOUS IN-
           FLUENT CONCENTRATIONS IN AIR AT 30°C
  Influent
Concentration          Pounds of Charcoal per 100 cfm Air
    (PPM)                90%          95%          98%

     10                  69           81            93

     20                  75           87           100

     28                  78           90           102

     35                  81           93           105
                               81

-------
                               REFERENCES
 1.   Vermeulen,  T.,  "Separation by  Adsorption Methods,"  Vol.  II Advances
     in Chemical Engineering,  edited  by T.  B. Drew  and J. W.  Hoopes,
     Academic  Press,  New York  (1958).

 2.   Altshuller, A.  P.  and  Cohen, I.  R.,  "Application of Diffusion  Cells
     to the Production  of Known Concentrations  of Gaseous Hydrocarbons,"
     Analytical  Chemistry,  32  802  (1960).

 3.   Barnebey, H. L., "Activated Charcoal in the Petrochemical Industry,"
     Chemical  Engineering Progress, 6J_ (1971).

 4.   Wheeler,  A., "Reaction Rates and Selectivity in Catalyst Pores,"
     Advances  in Catalysis, ^  250  (1951).

 5.   Manufacturer's  Technical  Data  for Pittsburgh Activated Carbon, The
     Calgon Corporation.

 6.   Wilke, C. R., and  Hougen, 0. A.,  "Mass Transfer in  the Flow  of Gases
     through Granular Solids Extended to  Low Modified Reynolds Numbers,"
     Transactions American  Institute  of  Chemical Engineers, 41, 445 (1945).

 7.   Grant, R. J., Manes, M.,  and  Smith,  S. B., "Adsorption of Normal
     Paraffins and Sulfur Compounds on Activated Carbon," American  Institute
     of Chemical Engineers  Journal, £, 403 (1962).

 8.   Hougen, 0.  A.,  and Watson, K.  M., "Chemical Process Principles,"  Part
     three, John Wiley  and  Sons, New  York, 1946.

 9.   Wertheim, E., "Textbook of Organic  Chemistry," Blakiston, Philadelphia,
     1945.

10.   Reid, E.  E., "Organic  Chemistry  of  Bivalent Sulfur," Vol. 1, Chemical
     Publishing  Co., 1958.
                                       82

-------
                              NOMENCLATURE
a         Surface area per unit volume of packed-bed
A         Quantity of adsorbent
b         Constant
C         Concentration
C         Initial concentration
 o
D         Distribution ratio, defined by equation 56
D         Diffusivity
dp        Particle diameter
D         Particle diffusivity
 P
D         Pore diffusivity
 pore
F         Feed rate
F.        Feed rate of component A
 Ao
k         Kinetic rate constant
kf        Fluid-phase mass transfer coefficient
k         Particle-phase mass transfer coefficient
 P
k         Pore mass transfer coefficient
 pore
L         Length
M         Molecular weight
N         Number of fluid-phase mass transfer units
N         Number of particle-phase mass  transfer units
 P
N         Number of pore mass transfer units
 pore
N         Number of reactor units
 R
p         Partial pressure
P         Total pressure
p         Upper limit of fluid-phase pressure
q         Concentration of adsorbate in  solid phase
q         Average concentration of adsorbate in solid phase
q         Maximum concentration of adsorbate in solid phase
q*        Concentration of adsorbate in  solid phase  in equilibrium with initial
 °        fluid phase adsorbate concentration
r         Radius
r         Rate of chemical reaction
                                       83

-------
                        NOMENCLATURE (Continued)
R         Separation factor, defined by equation 15
r         Internal particle radius
r         External particle radius
_P
r         Average pore radius
R         Gas Law Constant
 8
S         Cross sectional area
t         Time
T         Absolute temperature
v         Bed volume
V         Volume of fluid entering bed
V         Volume of adsorption system
 S
W         Mass of packed bed
X         Dimensionless fluid-phase concentration
X*        Dimensionless fluid-phase concentration in equilibrium with solid
          phase
x         Conversion of component A in a chemical reaction
 A.
x         Internal void fraction of particle
Y         Dimensionless solid-phase concentration
Y*        Dimensionless solid-phase concentration in equilibrium with fluid
          phase
Z         Through-put parameter, defined by equation 5
a         Constant in equation 21
a         Constant in equation 35
B         Correction factor for addition of resistances
e         Void fraction of packed bed
p.         Bulk density of packed bed
                                       84

-------
                                 TECHNICAL REPORT DATA
                          (Please read /nsl/iictiotis on the reverse before completing)
 \ REPORT NO
  EPA-650/2-74-084
                            2.
 4 TITLE AND SUBTITLE

 Odor Removal from Air by Adsorption on Charcoal
 7 AUTHOR(S)

 B. G.  Kyle and N. D. Eckhoff
                                    3 RECIPIENT'S ACCESSION NO


                                    5 HEPORT~DATE
                                    September 1974
                                                       e PERFORMING ORGANIZATION CODE
                                   8 PERFORMING ORGANIZATION REPOH1 NO
 9 PERFORMING ORG •\NIZATION NAME AND ADDRESS
 Kansas State University
 Manhattan, Kansas  66502
                                    10 PROGRAM ELEMENT NO.
                                    1AB015; ROAP 21AXM-061
                                    11 CONTRACT/GRANT NO

                                    EHSD 71-4
 12 SPONSORING AGENCY NAME AND ADDRESS

 EPA, Office of Research and Development
 NERC-RTP, Control Systems Laboratory
 Research Triangle Park, NC 27711
                                    13. TYPE OF REPORT AND PERIOD COVERED
                                    Final; Through 12/31/73
                                    14 SPONSORING AGENCY CODE
 15 SUPPLEMENTARY NOTES
 16 ABSTRACT
          The report gives results of an evaluation of the efficacy of charcoal for
 removing odorous organic vapors from extremely low concentrations in air, at amb-
 ient conditions.  Two systems were studied in detail: acetaldehyde--Pittsburgh PCB
 charcoal, and ethylmercaptan--Pittsburgh BPL charcoal. Fixed-bed breakthrough
 data were taken for the acetaldehyde—Pittsburgh PCB charcoal system at an acetal-
 dehyde concentration of 28 ppm.  These data were analyzed using a  modeling approach
 that indicated that the rate-controlling step was intraparticle diffusion. Preliminary
 design calculations based on this work indicate that fixed-bed adsorption with char-
 coal is a feasible process for the 98% removal of small concentrations of acetalde-
 hyde from polluted air. An integral  reactor approach using Pittsburgh BPL charcoal
 was used  in the study of the catalytic oxidation of ethylmercaptan in the concentration
 range 20-125 ppm. Diethyldisulfide was identified as a reaction product and the reac-
 tion was found to be one-half order in ethylmercaptan concentration. Preliminary
 design calculations show that 99% removal of ethylmercaptan from  air by catalytic
 oxidation  on charcoal appears to be  feasible.
17
                             KEY WORDS AND DOCUMENT ANALYSIS
                 DESCRIPTORS
                                          b IDENTIFIERS/OPEN ENDED TERMS
                                                                   C. COSATI I Icld/Group
 Air Pollution
 Odor Control
 Adsorption
 Charcoal
 Activated Carbon
 Acetaldehyde
Thiols
Air Pollution Control
Stationary Sources
Catalytic Oxidation
Organic Vapors
Ethylmercaptan
13B
                                                11G
                                                07C
18 DISTRIBUTION STATEMENT
                                           19 SECURITY CLASS (This Report)
                                           Unclassified
                                                                   21
                                                  NO OF PAGES

                                                     100
 Unlimited
                       20 SECURITY CLASS /Thupage)
                       Unclassified
                                                                   22 PRICE
EPA Form 2220-1 (9-73)
                                          84a

-------
APPENDIX
       85

-------
                500
ipoo
1,500
2.OOO
                                                                 TIME, min.
FIG.(A-I). APC-14, FIXED-BED TEST WITH 14 MESH CHARCOAL, 8.66 cm DEEP BED, AND 16.67  C9sec
          FLOW  RATE COMPARED  WITH MODEL BREAKTHROUGH CURVES

-------
 28
 24
 20
  16
  12
(T

u

le
SGLJD DIFFUSION


     ^	-
     EXPERIMENTAL
   0                   500                  1,000                1,500                2,OOO
                                                                     TIME, m!n.

      FIG.(A-2). APC-18, FIXED-BED TEST WITH 14 MESH CHARCOAL, 10.16cm DEEP BED, AND 23.75 C9feec

               FLOW RATE   COMPARED  WITH MODEL  BREAKTHROUGH  CURVES

-------
00

OC
       28
       24
       20
      E
      o.
      a.
       16
IU
2
o
        8
        4
                                                    ^--'  DIFFUSION
                                                     SOLID
                            500
                                           1,000
1,500
                                                                          TIME, min.
2 POO
           FIG.(A-3}. APC-19, FIXED-BED TEST WITH 14 MESH CHARCOAL, 13.82cm DEEP  BED AND 32.5 c<*>ec

                     FLOW  RATE  COMPARED  WITH MODEL BREAKTHROUGH  CURVES

-------
oc
       0                  500                 1,000                I,5OO                2,000
                                                                          TIME, min.

          FIG.lA-4). APC-20, FIXED-BED TEST  WITH 14 MESH CHARCOAL,  11.33 cm DEEP BED  AND  32.5
                           FLOW  RATE  COMPARED  WITH  MODEL BREAKTHROUGH  CURVES

-------
VO
o
     28
     24
      20
    i
            REACTION     /
              KINETICS ^^
.-'"SOLID DIFFUSION

              "EXPERIMENTAL
                                              1,000
                                I.5OO
                                                                         TIME,min.
2,000
          FIG.IA-5). APC-21,  FIXED-BED TEST WITH 14 MESH CHARCOAL, "7.56 cm DEEP BED, AND 32.5

                           FLOW  RATE  COMPARED WITH  MODEL BREAKTHROUGH CURVES

-------
     28
     24
     20
      12
    UJ
    (J
s   8
      8
      4
                                 REACTION KINETI
  ^ _ _ - -—' "SOLID" DIFFU SION

**"
                                  tE DIFFUSION
                           500
I.OOO
   1,500
                                                                            TIME, min.
2.OOO
          FfG.(A-6). APC-22, FIXED-BED  TEST WITH 14 MESH CHARCOAL,  12.59cm DEEP  BED, AND  23.75

                           FLOW  RATE  COMPARED WITH  MODEL  BREAKTHROUGH  CURVES

-------
  28
  24
  20
I16
a
LJ

U
  12
  8
              REACTION

                KINETICS
                       DFFUSION
                       500
1,000
1,500
2,OOO
                                                                     TIME, min,


       R6.(A-7). ARC-26, FIXED-BED TEST WITH  14  MESH CHARCOAL, 20.32 cm DEEP BED, AND 78.33 c*sec

                        FLOW  RATE COMPARED  WITH MODEL BREAKTHROUGH  CURVES

-------
  28
  24
lie
&
 •»
o

I 12


u
  8
    K

    t
                    PORE DIFFUSION
                            I	I	I
          1,000
1,500
2,OOO
2,500
                                                                           TIME, min.
3.OOO
      FIG.(A-S). APC-29, FIXED-BED TEST WITH  14 MESH  CHARCOAL,   45.72 cm DEEP BED AND 32.5
                        FLOW RATE  COMPARED WITH  MODEL BREAKTHROUGH  CURVES

-------
vC
                                REACTION KINETICS
         0                 500                 I.OOO                 1,500                 2,000
                                                                          TIME, min.

           FIG.(A-9).  APC-30,  FIXED-BED TEST WITH  14 MESH CHARCOAL, 42.72 cm DEEP BED, AND 49.5 c?sec
                            FLOW RATE  COMPARED  WITH MODEL BREAKTHROUGH CURVES

-------
F
       24
        12
     uj
5    8
        8
         1
                    REACTION KINETICS
   DIFFUSION

EXPERIMENTAL
                             PORE DIFFU90N
                                                           J	1	L.
                             500
                                        I,OOO
                            1,500
                                                                            TIME, min.
2,000
            FIG.(A-IO). APC-31, FIXED-BED  TEST  WITH 14 MESH  CHARCOAL, 45.72 cm DEEP BED AND 69L22 C9sec
                      FLOW RATE  COMPARED  WITH MODEL BREAKTHROUGH CURVES

-------
                                  ..'•SOLID DIFFUSION


                                         'EXPERIMENTAL
0                   500                 I.OOO                 1,500                 2.OOO
                                                                    TIME, mi n.
  FIG.(A-II).  APC-32, FIXED-BED TEST WITH 14 MESH  CHARCOAL, 45.72 cm DEEP BED  AND 59.4 C9sec
             FLOW RATE  COMPARED  WITH MODEL BREAKTHROUGH  CURVES

-------
     5OO
1,000
1,500
2,000
2,500
                                                                      TIME, min.
FIG.(A-I2). ARC-38 .FIXED-BED TEST  WITH 10 MESH CHARCOAL,  45.72 cm  DEEP BED  AND 7935
         FLOW RATE COMPARED WITH  MODEL BREAKTHROUGH CURVES

-------
10
00
                                  I POO
,500
2.OOO
2.5OO
                                                                                TME.min.

          RG.(A-I3).  APC-39,  FIXED-BED TEST  WITH 12 MESH  CHARCOAL,  45.72 cm  DEEP BED,  AND 45S3

                    FLOW RATE  COMPARED  WITH MODEL  BREAKTHROUGH  CURVES

-------
          FIG.(A-I4)   GAS  CHROMATOGRAPH CALIBRATION  CURVE.
  ISO
o

w 100
u
a.
O 50
c

g
            10      20      30      40      60


                     PPM  OF ETHYLMERCAPTAN
60
70
                                 99

-------
   too
u
I
o
   80
   60
   40
   20
             10
   20      30      40      50      60
ACETALDEHYDE  CONCENTRATION, ppm
                          ATENUATION =2
                          FLAME RANGED O.I
        FKMA-15). CHROMATOGRAPHIC  CALIBRATION  CURVE
                  ACETALDEHYDE
                               100

-------