You are on page 1of 35

KINETIC MODELING STUDY OF

CARBON NANOTUBES
SYNTHESIS BY FLUIDIZED BED
CHEMICAL VAPOR DEPOSITION
ARTICLE CRITICAL REVIEW

ANDRS FELIPE MUOZ PATIO

REACTOR ANALYSIS AND DESIGN


THE AUTHORS

R. Philippe LCC/ENSIACET/INPT, UPR CNRS 8241, Toulouse University, 31077


Toulouse Cedex 4, France ARKEMA LACQ Research Center, 64170 Lacq, France
P. Serp and Ph. Kalck LCC/ENSIACET/INPT, UPR CNRS 8241, Toulouse University,
31077 Toulouse Cedex 4, France
S. Borde`re, D. Plee, and P. Gaillard ARKEMA LACQ Research Center, 64170 Lacq,
France D. Bernard ARKEMA, 92705 Colombes Cedex, France
B. Caussat LGC/ENSIACET/INPT, UMR CNRS 5503, 31106 Toulouse Cedex 1, France
ABOUT THE PAPER

Journal: American Institute of Chemical


Engineers AIChE J.
Vol: 55
Pages: 465-474
Month: February
Year: 2009
STATEMENT OF THE PROBLEM

The authors present an implementation of the kinetic and physical laws developed in a
previous study; in a modified version of the bubbling bed Kato and Wen model to
represent multiwalled carbon nanotubes (MWCNTs) synthesis by catalytic chemical
vapor deposition from ethylene as carbon source and using an Fe/Al2O3 catalyst. In
addition, study the influence of the main operating parameters on the evolutions with
time of the species molar fractions, the weight of MWCNTs formed, and the bed
characteristics; can help designing new reactors.
NOTATION
RESULTS FROM PREVIOUS WORK / APARENT
KINETIC EQUATIONS
The chemical mechanism used to describe the decomposition of ethylene and the
formation of MWCNTs is composed of three apparent catalytic reactions:
RESULTS FROM PREVIOUS WORK / APARENT
KINETIC EQUATIONS
The kinetics laws for each reaction presented before is like follow:

Between 550 and 650 C:

Between 651 and 750C:


RESULTS FROM PREVIOUS WORK / PHYSICAL LAWS
ABOUT THE EVOLUTION OF THE BED
In this point the paper present some problems, thus:
The paper say that In the rst part of the study, we have determined the evolution of the
main physical parameters of the FB, i.e., mean volume diameter of powders, apparent
grain density, xed bed height, and minimum uidization velocity, with run duration. This
analysis isnt in the part 1 mentioned.
The paper say that It appeared that the discrepancy between these two sets of values
was in average of 15%. As a consequence, the model presents an intrinsic error of about
15% due to these physical laws of evolution. this affirmation is not clear and is been
responsible of the 15% of the difference between the model and the experimental data.
RESULTS FROM PREVIOUS
WORK / PHYSICAL LAWS
ABOUT THE EVOLUTION
OF THE BED

Physical laws of evolution of


the fluidized bed and of the
particles features with X1 the
MWCNTs productivity. (a)
Fixed bed height, (b) average
diameter of particles, (c)
apparent density, and (d)
minimum fluidization gas
velocity.
RESULTS FROM PREVIOUS WORK / PHYSICAL LAWS
ABOUT THE EVOLUTION OF THE BED
Analysing the data the Authors say that the evolutions of dp and qp lead to a decrease of
the minimum uidization velocity between 0 and 0.4 gC/gcata and then to an increase of
Umf until values higher than its initial value, for 2.7 gC/gcata.
This is just an observation from the graphs, but could be necessary to understand the
behavior, to relate the Umf with the bed height.
PRESENTATION OF THE
MODEL
Representation of a fluidized bed according
to Kato and Wen, work Blubble assemblage
model for fluidized bed catalytic reactor.
THE ASSUMPTIONS

The fluidized bed is isothermal.


It is represented by a series of compartments whose height is equal to the mean
diameter of bubbles at the corresponding level.
THE ASSUMPTIONS

Bubbles are growing by coalescence as they ascend through the bed; the bubble diameter is
given by the empirical relation of Kobayashi:

Where

The number of holes of the distributor per surface area of column b0 is equal to 30 cm-2 for
the analyzed case.
THE ASSUMPTIONS

Each bubble is surrounded by a cloud of particles; its volume is calculated by the relation
of Davidson and Harrison.

Where
THE ASSUMPTIONS

Each compartment is constituted of two phases, the bubble one corresponding to the
bubbles and their cloud, and the emulsion one, in which wakes are included; the volume
of the bubbles for the kth compartment is given by:

Where

And the resulting volume of the emulsion for the kth compartment is given by:
THE ASSUMPTIONS

Each phase is supposed to be perfectly mixed.


The convective mass transfer in the emulsion phase is neglected.
In each compartment, mass transfer occurs between the two phases and the exchange
coefficient is given by the Kobayashi correlation:
THE ASSUMPTIONS

A final assumption was made about the slugging regime, and was, that it is reached when
the bubble diameter is equal or higher than a limit value corresponding either to the
diameter given by the relation of Grace, if this value is lower than 0.8 times the reactor
diameter.
When the slugging regime occurred, the authors arbitrarily fixed the bubble diameter to
this value to perform the calculations, but the final results were clearly less precise.
MODIFFICATION TO THE MODEL OF KATO AND
WEN
Considering the stoichiometric of the reactions, the authors have modified the mass
balance equations of the model to consider the influence of variations of gas flow rate
induced by chemical reactions, and say that; Indeed, experimentally a nonnegligible
increase of the total flow rate has been observed by Gas Chromatography during
MWCNT synthesis.
MODIFFICATION TO THE MODEL OF KATO AND
WEN
The modified mass balances equations are the following:
For the bubble phase:

For the emulsion phase:


MODIFFICATION TO THE MODEL OF KATO AND
WEN
Where the parameter qEB quantifies the gaseous flux transferred from the emulsion to
the bubble phase, per unit volume of bubble and is equal to:
MODIFFICATION TO THE MODEL OF KATO AND
WEN
Which allows to obtain the evolution of the superficial gas velocity U from a
compartment to another, thus:
RESULTS AND
DISCUSSION
Experimental and Simulated
Results
RESULTS AND DISCUSSION

Comparison between experimental and simulated


productivities (runs Cin1 to Cin25). (a) X1
MWCNTs productivity, (b) X2 ethane
productivity, and (c) X3 methane productivity.
RESULTS AND
DISCUSSION
Comparison between
experimental and simulated
(a) fixed bed height and (b)
average particle diameter.
RESULTS AND
DISCUSSION
After, the authors present the spatial
and temporal evolutions of some
process parameters, thus:
Axial evolution of ethylene and
hydrogen molar fractions in each
phase for various run durations
under the conditions of run Cin3. (a)
Ethylene molar fraction in the bubble
phase, (b) hydrogen molar fraction in
the bubble phase, (c) ethylene molar
fraction in the emulsion phase, and
(d) hydrogen molar fraction in the
emulsion phase.
RESULTS AND
DISCUSSION
Simulated evolution of (a) the
fluidization ratio U0/Umf and
(b) the expanded bed height
with run duration for various
temperatures.
Simulated evolution of
MWCNTs weight with run
duration for various
temperatures.
RESULTS AND
DISCUSSION
Influence of the inlet molar
fraction of ethylene on (a) the
ethylene conversion rate and
(b) the final mass of
MWCNTs for various
temperatures.
RESULTS AND
DISCUSSION
Influence of the inlet molar
fraction of ethylene on (a) the
final expanded bed height and
(b) the mean grain diameter
for various temperatures.
RESULTS AND
DISCUSSION
Simulation of Various Process
Scale Up.
It is worth noting that the
slugging regime never occurred
for these simulations since the
ratio between the final fixed
bed height and the reactor
diameter H/D never exceeded
3.2. This strengthens the validity
of these predictions.
CONCLUSIONS OF THE ARTICLE

A model of fluidized bed has been developed to represent MWCNTs synthesis by Catalytic Chemical Vapor
Deposition from ethylene as carbon source and using an Fe/Al2O3 catalyst. The kinetic and physical laws
developed in Part 1 have thus been implemented in a modified version of the Kato and Wen model.
To the best of our knowledge, this is the first time that a two-phase bubbling bed model is used to simulate a
viable process of large scale MWCNTs production.
Its domain of validity is of course restricted to the operating range used to develop the associated kinetic and
physical laws. Its accuracy is also limited by the fact that the kinetic laws are not intrinsic and do not account
for any deactivation phenomenon.
The absolute deviation for MWCNT productivity between these experimental results and the simulations was
of 20.8% for the whole data and of 17.3% when considering only experiments for which the bed was mainly in
bubbling regime.
CONCLUSIONS OF THE ARTICLE

A perspective of the work could then be to implement these kinetic and physical laws in a
slugging bed model11 or in an Eulerian multifluid model-like MFIX12 to better represent the
hydrodynamics of the fluidized bed in particular in slugging regime and the associated
interphase mass transfers.
The code has also allowed to calculate the influence of the main operating parameters on the
evolutions of the species molar fractions and on the bed characteristics vs. run duration.
This numerical tool can then help to better control the evolution with time of the process
behavior, for instance, in terms of fluidization regime, mean diameter of grains, or expanded
bed height. In a last part, the influence of the bed features as the reactor diameter and weight
of catalyst has also been studied proving the interest of such model for scale up.
CONCLUSIONS OF THE ARTICLE

Simulations have shown that the productivity in MWCNTs could reach 8 kg/h in a 45-cm
diameter reactor operating in semi-batch conditions. This corresponds to around 74
tons/year of MWCNTs selectively produced.
An increase of reactor diameter provides better interphase contacts due to lower bubble
diameters, relatively to reactor diameter, and then enhances the process productivity in
MWCNTs and its selectivity regarding ethane and methane. The interest of the process
to massively produce MWCNTs is then fully demonstrated.
GENERAL DISCUSSION ABOUT THE ARTICLE

Analyzing the results and the conclusions presented by the authors, I can say that in the
article there is coherency and the information presented were evaluated carefully; there
are some point in which could be missing information, but all the analyses seems to be
ok.
The work was focused on the comparison between the modeling and the experimental
data, which correspond with the purpose of the article, but could have present and take
advantage of the data to present more details about the phenomena; or use it to explain,
in a more detailed way, the behavior of some variables.
GENERAL DISCUSSION ABOUT THE ARTICLE

The title of the article was appropriate and clear about the topics on it, the abstract was
specific, representative of the article, and was presented in the correct form; and with the
introduction let know to the reader the purpose of the work.
I consider that whole the discussion raised in this paper, was relevant and very important
for the development of the technique of mass production of multiwall carbon nanotubes,
like a very important material for the futures developments.
THANKS
ANDRS FELIPE MUOZ PATIO