STUDY OF ACETYLSALICYLIC ACID ADSORPTION BY

MOLECULAR SIEVES
G. A. Assumpo1, J. G. R. Poo2, R. Fernndez-Felisbino1, D. Cardoso3, E. L.
Gomes1,* (1) Department of Exact and Earth Sciences, Chemical Engineering Sector, Federal
University of So Paulo (UNIFESP), R. Artur Riedel, 275, Diadema SP, 09972-270, Brazil.
(2) Department of Chemical Engineering, Faculty of Industrial Engineering University Center
(FEI), Av. Humberto Alencar Castelo Branco, 3972 So Bernardo do Campo SP, 09850901, Brazil. (3) Department of Chemical Engineering, Federal University of So Carlos
(UFSCar), Rod. Washington Luis, Km 235 So Carlos SP, 13565-905, Brazil.
*Author to whom all correspondence should be addressed. E-mail: eliezer.ladeia@unifesp.br
(E. L. Gomes). Phone: +55 11 3319-3545 Fax: +55 11 3319-3400.

ABSTRACT: Molecular sieves can be used as a matrix host to encapsulate several drugs. Here
we studied the adsorption of acetylsalicylic acid (ASA) in HUS-Y, HZSM-5 and silicalite-1
zeolites, and in the mesoporous sieve Si-MCM-41, in order to compare their adsorption
capacities. The materials used presented generally good performance for ASA adsorption.
Among the zeolites, HZSM-5 adsorbed relatively more drug than the others. The Si-MCM-41
samples synthesized with fumed silica presented better performance for adsorption than those
ones synthesized with TEOS. The best mesoporous sample for adsorption, synthesized with
silica fumed, was chosen for the kinetic assay. The adsorption kinetics of ASA for this sample
fitted well to a model of second order. The initial loading rate and diffusion coefficient for the
ASA-(mesoporous sieve) system were 0.153% (mg ASA)/(mg solid.min) and 5.02 x 10-17 m2/s,
respectively. The materials were characterized by XRD, SEM and TEM, N2 physisorption and
infrared spectroscopy.

1. INTRODUCTION
Recently the use of inorganic matrices associated to drugs, in order to modify its
adsorption or release in the human body or even improve their adsorption from environmental,
is a very common theme in the area of pharmaceutical sciences (Rakic et al. 2013). Studies
involving several drugs have been reported, such as ibuprofen (Horcajada et al. 2004; Gonzalez
et al. 2013; Charnay et al. 2004), acetylsalicylic acid (Beninati et al. 2008), and cancerigen
agents (Zhou and Zhu, 2005), all associated to inorganic matrices such as mesoporous sieves,
activated carbon or clays.
Among the drug delivery systems, porous inorganic materials are emerging as a new
category of systems "host/guest" (Moller and Bein, 1998): for example, we can mention
synthetic zeolites, silica xerogel and porous ceramics (Unger et al. 1983; Uglea et al, 1994). The
advantages of zeolites and mesoporous silicas for biomedical applications include
biocompatibility, low toxicity, large surface areas and the ability to control the physicochemical

properties (Charnay et al. 2004). Due to these interesting features, there is a significant and
growing interest in these potential supports.
The incorporation of active substances in inorganic materials has many advantages such
as reducing the risk of dosage errors, diminution of the necessary number of administrations and
avoiding allergic reactions (Amos et al. 1971), besides of a better use of the active compounds
located in the digestive tract, providing greater convenience and comfort to patients (Nema et
al. 1997).
Zeolites were discovered by Swedish mineralogist Cronstedt in 1756. Their selective
adsorption properties led to the known term "molecular sieve" which describe this class of
materials that have the ability to separate molecules (or adsorb them from a solution) based on
its size (Giannetto 1990). However, zeolites have a diffusional limitation for some applications
due to the diameter of their micropores (Rakic et al. 2013).
In the 90s, researchers at Mobil Co. patented the materials of M41S family (MCM-41,
MCM-48 and MCM-50), with well-defined pore diameters in the range of mesopores and high
surface area (> 1000 m2/g) (Beck et al. 1992). The larger pore diameter enables the diffusion of
bulky molecules to their interior. MCM-41, for example, has been extensively studied and
allows the occlusion of several kinds of drugs into their pores, offering an alternative as a drug
delivery system (Mollo et al. 1997; Vallet-Regi et al. 2001). In contrast to some minerals that
have disordered pores with a wide range of sizes and irregular shapes, MCM-41 exhibits
excellent structural characteristics, such as high surface area and cylindrical ordered mesopores
with a narrow diameter distribution (Cheng et al. 1997). By the appropriate choice of surfactant
or by inserting swelling agents which increase the size of their micelles (Beck et al. 1992; Blin
et al. 2002), these mesopores can be obtained in a range from 20 to 100 . Moreover, it is
possible to chemically modify the mesopores surface through the "grafting" process, that is,
anchoring organic functional groups on them (Brunel et al., 2000; Derrien et al. 1998 ). Another
advantage is that fumed silica is widely used as a pharmaceutical excipient. Thus, it would be
interesting to use mesoporous silica formulations for the controlled adsorption and release of
drugs.

Acetylsalicylic acid (ASA), also called 2-acetoxybenzoic acid (Figure 1) was first
prepared in 1893 by Charles Frederic Gerhardt by reaction of acetyl chloride with salicylic acid
(Beninati 2008). In 1898, the Bayer Co. patented this drug under the name Aspirin. It is one of
the most successful synthetic compounds in the field of pharmaceutical chemistry and it is
employed as an analgesic for pain of low intensity, with mild sedative action and excellent
antipyretic action. Additionally, the anticoagulant effect of aspirin has been widely studied
(Watson and Berquist 2000).

INSERT FIGURE 1

This paper presents an exploratory study of the adsorption of acetylsalicylic acid on

molecular sieves HUS-Y, HZSM-5, silicalite-1 and Si-MCM-41, and had as main objectives
relate the textural properties of micro- and mesoporous sieves with their adsorption capacities,
determine the adsorption kinetics and estimate the diffusion coefficient for the
AAS/mesoporous molecular sieve.

2. EXPERIMENTAL
2.1. Molecular sieves
The zeolites used in these experiments, microporous molecular sieves in the acid form,
were previously characterized with respect to their elemental composition, specific surface area
and micropore volume (Table 1). The silicalite-1 sample was synthesized according to the
method proposed by Prakash et al. (1998), using tetrapropylammonium hydroxide as structural
template and tetraethyl ortosilicate (TEOS).

INSERT TABLE 1

The mesoporous adsorbents used in these experiments were Si-MCM-41 molecular

sieves. Samples T1, T2 and T3 were prepared based on a modification of the methodology

proposed by Cai et al. (1999), synthesized at 100 oC during 2 days, using tetraethyl ortosilicate
(TEOS, Aldrich, 98%) as silica source, hexadecyltrimethylammonium bromide (CTABr, Sigma,
98%) and amoonium hydroxide (Cintica Qumica, 25% w/w sol.), with the composition
1 SiO2: 0.15 CTABr: 19.5 NH3: 82.5 H2O. The samples A1, A2 and A3 were prepared
according to the method proposed by Cheng et al. (1997), synthesized at 150 oC during 2, 4 and
10 days using fumed silica (Aerosil 380, Evonik, 99.99 %) as silica source,
tetramehylammonium hydroxide (TMAOH, Aldrich, 97%) with the composition 1 SiO2:
0.27 CTABr: 0.19 TMAOH: 40 H2O.

2.2. Characterization
A. Small Angles X-ray Diffraction. The diffractograms were obtained from a diffractometer
Bruker, D8 Advanced model, employing CuK radiation and nickel filter, operating at 40 kV
and 40 mA with goniometer speed of 1/min. The samples were scanned in the range 2 from 1
to 10o.
B. N2 Physisorption. The physisorption isotherms were obtained using a Quantachrome
Corporation (Nova-1200e) equipment. Before each analysis, about 0,05 g of the sample was
treated at temperature of 150 C for 2 h under vacuum.
C. Thermogravimetric analyses (TG). A thermobalance DTG-60H da Shimadzu was employed,
using 10 mg of the sample, under synthetic air flow, from 25 to 600 C, with a temperature
heating rate of 10 C/min.
D. Scannig Electron Microscopy. The analyses were performed in a scanning electron
microscope Philips, XL30-FEG model, operating at 30 kV, with 2 nm resolution and maximum
magnification of 1.000.000 X. Some samples were also analyzed in a FEI Company SEM,
Quanta 3D FEG model, operating at 30 kV, and magnification from 1500 to 100.000 X.
E. Transmission Electron Microscopy. A TEM Philips CM-120 with 0,2 nm resolution and
maximum magnification of 660.000 X, was used operating at 120 kV.

2.3. Acetilsalicilic acid (ASA) adsorption measurements

2.3.1. Total adsorption capacity

This experiment aimed to fully load the drug in the molecular sieve in a period of 24 h.
First, 150 mg of a calcined sample, previously activated for 24 hours at 120 C, was weighed.
After that, the solid was added to a solution of acetylsalicylic acid (1 mg/mL ethanol) under
stirring (Charnay et al. 2004). After 24 h, the suspension was centrifuged; the solid obtained
was dried for 24 hours at 110 C and finally submitted to thermal analysis.

2.3.2. Adsorption kinetics

This experiment was carried out aiming to follow the loading of the molecular sieve
during 4 h, taking aliquots of 5 mL from the suspension, at predetermined time intervals. The
aliquots were centrifuged and the solid was dried for 24 hours at 110 C and, after that,
submitted to thermal analysis.

3. RESULTS AND DISCUSSION

3.1. Zeolites and Si-MCM-41
The X-ray diffractograms (XRD) of HZSM-5 zeolites, HUS-Y and Silicalite-1 (not
shown here) proved the formation of the desired phases, showing their characteristic diffraction
peaks and presenting good crystallinity, without the presence of undesirable phases.
In Figure 2 the diffractograms of Si-MCM-41 synthesized with TEOS are presented.
Comparing these XRD patterns with bibliography data (Meynen and Cool 2005) it can be said
that these materials correspond to the desired product and there was no formation of other
phases. For all samples (Tn) is observed the peak related to the plan (100), more intense. The
peaks related to secondary plans (110) and (200) are present in the samples T2 and T3, and less
obvious, in the sample T1.

INSERT FIGURE 2

Regarding to the structural organization of these samples Tn, we can observe that the
peak (100) of samples T2 and T3 are narrower and more intense, while the peak (100) of T1
sample is wider and less intense. According to Brinker (1996), we could say that, starting from
the material less organized for the more organized, we would have the sequence T1<T3<T2
based only on this parameter.
In Figure 3 the diffractograms of Si-MCM-41 samples synthesized with Aerosil 380, with
different thermal treatment times, are presented. The materials (An) obtained also correspond to
the Si-MCM-41 without parallel phases, despite of the A1 sample being very disorganized. For
these samples we observe the peak (100) more intense (A2 and A3) and almost imperceptible
for A1. The secondary peaks (110) and (200) are present only for sample A2.

INSERT FIGURE 3

Regarding to the structural organization of these samples An, the sample A2 presents a
peak (100) narrower and more intense, while the sample A3 presents the peak (100) wider and
less intense. We would have the sequence of increasing organization as A1<A3<A2. The
thermal treatment time effect for these samples is evident in these diffractograms: with only 2
days there is no sufficient time for the formation of the Si-MCM-41; with 4 days the material is
more organized; with 10 days the phase begins to dismount and disorganize.
In Table 2 we have the textural parameters for the Si-MCM-41 samples obtained from the
diffractograms. The LMA organization parameter is consistent with the previous discussion.
However, when we analyze only the parameter organization degree (Go), it is not possible do a
good comparison in this sense since the areas under the peaks (100) depend heavily on the
amount of organized or disorganized pores.

INSERT TABLE 2

A parameter that could minimize this distortion would be Go = Go/LMA that takes into
account the full width at half maximum (FWHM). Thus, analyzing the values of Go, FWHM
and Go, we could say that the samples A2, T2, A3 and T3 are well organized, with hexagonally
structured pores. Sample A1 shows no arranged pores and sample T1 shows a lot of less
organized pores. In Table 2 we also have the hexagonal framework parameter (aH), ranging
between 39 and 66 . These values are consistent with those found by other authors (Meinen
and Cool 2005; BECK et al. 1992).
In Figure 4a is presented the Si-MCM-41 SEM micrograph (sample A2) synthesized with
Aerosil 380 (thermal treatment: 4 days at 150 oC). It is observed that it consists of irregularlyshaped clusters ranging between 5 and 15 m. A more careful analysis shows that these clusters
are formed of small crystallites of ~ 0.1 m. Carvalho (1997), when synthesizing Si-MCM-41
with fumed silica, observed in the same sample irregular particles and others with hexagonal
section, claiming that irregular particles indicated an incomplete "crystallization" of the reaction
gel, justifying the low specific areas and the diffraction patterns with a low degree of
organization achieved.
Figure 4b shows a SEM micrograph of Si-MCM-41 synthesized with TEOS (sample T2).
The synthesis with TEOS produces particles in geometries that ranging from globules to rods,
with sizes between 1 and 10 m. This morphology is related to monomeric silica (TEOS) that
facilitates the production of non-agglomerated particles and with a tendency to form hexagonal
particles.

INSERT FIGURE 4a

INSERT FIGURE 4b

In Figure 5 shows a TEM micrograph in bright field of sample A2 (Si MCM-41, Aerosil).
We can observe a pattern of parallel lines in the particle which correspond to pores seen along

direction [010], confirming the formation of mesopores. Analyzing the scale and considering a
measurement error, we have mesoporous diameter of approximately 2.6 nm.

INSERT FIGURE 5

3.2. ASA Adsorption

3.3.1. Total adsorption capacity
Figures 6 to 8 show the N2 physisorption isotherms of the representative samples for
each group of molecular sieves before and after loading of ASA for 24 h. Table 3 summarizes
the textural parameters obtained for each case.
The HZSM-5 zeolite (Figure 6), used for the adsorption of ASA, showed a behavior
similar to HUS-Y zeolite and Silicalite-1. They show type I isotherms as expected. The
maximum specific volume in physisorption of HZSM-5 original sample was 114 cm3/g, value
also compatible with those observed by Giannetto (1990). It is observed that the isotherm, even
after the ASA loading, presents N2 physisorption isotherms of type I, however shifted down, i.e.
the amount adsorbed was lower because the pores were filled by AAS and the solvent (ethanol).
From the data of Table 3 it can be said that, among the zeolites, Silicalite-1 adsorbed less
ASA and solvent (2.1 %) and that HZSM-5 showed the best performance (38.6 %) in relation to
the volume available for adsorption. HUS-Y zeolite also offers good performance (33.1 %) and
can be applied with good efficiency for this process.
It should be noted that the good performance of HZSM-5 and HUS-Y in relation to
Silicalite-1 can be possibly justified by the presence of aluminum in their structures, which
provides a charged framework that must be compensated, probably adsorbing ASA more easily
in these materials. Another aspect to take in account is the kinetic diameter of ASA (0.59 x 0.31
x 0.80 nm) that allows a better diffusion in HUS-Y pores than other zeolites, confirming its
good capacity adsorptive.

INSERT FIGURE 6

Figure 7 presents the physisorption isotherms of sample A2 (Si-MCM-41, Aerosil) used

for ASA adsorption during 24 h and that present similar behavior as sample A3. We can observe
type IV isotherms with H2-type hysteresis (slit-pores). The maximum specific physisorbed
volume for sample A2 was 541 cm3/g, value consistent with that is observed in bibliography.
Analyzing Table 3, the sample A1, which presented an isotherm with behavior next to the type
II (solid with few pores) and a diffractogram without peaks, showed that these pores were
efficient for the adsorption of ASA. The sample A3, despite being less organized, presented
good efficiency in adsorption. Possibly there is an influence of thermal treatment time on
particle size, because the sample A3, synthesized 10 days, provided a material with smaller
particle size, favoring the diffusion to the interior of the mesopores and increasing the ASA
sorption.
Based on the graphics and in Table 3 it can be said that the mesoporous material that
adsorbed less ASA and solvent was sample A2 (17.1%) and the best performance in this set was
the sample A3 (61.1 %) in relation to the volume available for adsorption. Despite the large
available volume, the sample A2 was not so efficient in ASA adsorption possibly because its
particle size is very large, with a long average free path to go, hindering the diffusion of ASA
and solvent molecules and not fully using the volume of the particle.

INSERT FIGURE 7

Figure 8 presents the physisorption isotherms for the sample T2 (Si-MCM-41),

synthesized with TEOS. The samples exhibit type IV isotherms with an unclear hysteresis. The
adsorbed volume was 470 cm3/g, the smallest among the Tn samples. Although the sample T1
presents smaller organization (FWHM), i.e, mesopores not ordered hexagonally, its volume is
significantly larger than the structurally organized samples T2 and T3. After loading of ASA,
we can observe that the isotherm is still of type IV, reproducing the behavior already discussed
for these kind of materials. Based on Table 3 it can be said that sample T1 was the material that

adsorbed less ASA and solvent (6.6%) and the sample T3 presented the best performance
(12.7%) in relation to the volume available for adsorption. Despite the large volume available,
the sample T1 was not effective in adsorption of ASA, possibly because their mesopores are not
arranged hexagonally, presenting tortuosity and irregular pores diameters, which difficults the
ASA and solvent molecules diffusion inside the pores.
The low loading of the sample T1 shows that the area and the volume are not the only
parameters to be considered to perform the adsorption of ASA, since the organization should
also be considered. Once the kinetic diameter of ASA is much smaller than the pore diameter
(DP), diffusional restrictions are not important, but cannot be neglected. This diffusion would be
of Knudsen, where the molecules adsorption is still influenced by the pore dimension.

INSERT FIGURE 8

INSERT TABLE 3

Figure 9 presents the TG curves of samples An, representing the relative mass loss of
each sample. The lost mass contains ASA and solvent (ethanol). The TG curves for the other
samples are similar. Thus, for a better view of the data, Table 4 summarizes the data obtained.

INSERT FIGURE 9

INSERT TABLE 4

From Table 4, in relation to zeolites, HUS-Y presents the largest total mass loss (29,9%),
and consequently higher adsorbed mass of ASA (11,71 %). This result is in agreement with Datt
(2012), who found that the total loading of ASA on zeolite HUS-Y was 23.5%, considering
ASA and solvent. Among the samples An, sample A3 presented higher total mass loss (9,47 %),
while among the samples Tn, sample T1 was that with the higher total mass loss (10,98%).

From TG curves of the samples and standard TG of ASA, it was possible to identify how much
of the mass loss corresponds to the ASA and solvent.
Observing Table 4, although much of the lost mass is solvent, it is confirmed that the
HUS-Y zeolite was material that most adsorbed ASA (11.71 %). In contrast, Silicalite-1
adsorbed less ASA and solvent, confirming the result obtained by physisorption. However, the
HZSM-5, in this study, did not have the same result as the physisorption, because the accuracy
of this analysis allows this discrepancy. We also have confirmed that the largest diameter of the
pore HUS-Y justifies its higher adsorption capacity.
Analyzing the mesoporous molecular sieves in Table 4, it was confirmed that the samples
A3 and T1 were the ones that most adsorbed ASA (4.73 % and 5.24 %, respectively). Similar
explanation for the HZSM-5 low performance can be given with respect to the low value
observed for sample T3, in this analysis.
The solvent used in experiments adsorbs aplenty, interfering in the ASA adsorption
process, which is the compound that we wish to occlude in the molecular sieves. The solvent
represent up to 60 % of the adsorbed material in the case of zeolites and up to 67 % for
Si-MCM-41 samples.
At this point we can discuss the preferential adsorption of ASA in the molecular sieves
which have aluminum in their structures. The sieves that adsorbed more ASA were HUS-Y and
HZSM-5 (acid forms). Aspirin is more soluble in ethanol (0,06 mol ASA/mol ethanol) than in
water (0,00066 mol ASA/mol water) (Grunwald and Berkowitz 1951). Ethanol is a solvent less
polar than water and so is able to more easily dissolve low-polarity substances. Possibly the
interaction of ethanol with ASA generates a clathrate that is an inclusion compound in which
the guest molecule (ASA) is in a cage formed by the host molecule (EtOH) or by a lattice of
host molecules. This clathrate [ASA-EtOH]+ interacts with the negative molecular sieve lattice,
adsorbing both molecules and explaining the high amount of solvent adsorbed. The almost
neutral lattice without aluminum of Silicalite-1 and Si-MCM-41 generates few sites for this kind
of interaction (silanol groups), justifying their low performance.

3.3.2. Adsorption kinetics

In this study we aimed to obtain preliminary adsorption kinetics of ASA for a
mesoporous molecular sieve. The sample A3, with the highest adsorption capacity among the
Si-MCM-41 (Aerosil) synthesized was selected for this application and determination of kinetic
parameters. Another reason that led to the selection of this sample is that the cost of preparation,
involving fumed silica, is much lower than the samples synthesized with TEOS, enabling their
potential use in the pharmaceutical industry.
Table 5 presents the data of adsorption kinetics experiment for ASA at 36.5 C in a
period of 240 min, obtained by thermogravimetric analysis. The kinetic curve is shown in
Figure 10. The experimental data obtained show excellent consistency, because a nonlinear
exponential fit shows no significant deviations (R2 = 0.9994), which proves the good quality of
the analysis method. The loading was progressive until the relative mass of 7.5% of AAS. After
60 min, the loading has reached its maximum, becoming constant from that time.

INSERT FIGURE 10

INSERT TABLE 5

The time required to reach equilibrium was graphically estimated as 240 min. The data
were then adjusted mathematically for kinetic models of adsorption of first and second order.
Starting up from the general order n equation, represented by equation (1):

= . ( )

(1)

In equation (1), qe is the adsorption capacity in equilibrium, n is the order of the kinetic
model, qt is the sorption capacity at a defined time t and kn is the constant rate of adsorption.
For the 1st order model proposed by Lagergrem, n takes the unit value and equation (1)
becomes the equation (2):


= . ( )

(2)

Thus, the rate of temporal variation in adsorption capacity would be linearly proportional
to the gradient of the adsorption capacity. A linearization is given in equation (3):

log = 1 +

(3)

Equation (3) suggests that the relationship between log(qt-qe) and time is linear. However,
qe is unknown, and adopt the highest value of qt for this value leads to the elimination of an
experimental point because equation (3) is not defined at this point. Thus it was assumed that qe
would be 7.5%, corresponding to the longest time of analysis. The linearization and these
estimations were made for the experimental data obtained. From t and ln(qe-qt), k1 values were
calculated for each time, and subsequently using the average (k1m) of these values in the model
evaluation. In Table 6 the data obtained for the kinetic model of adsorption of first order are
presented.

INSERT TABLE 6

If we discard the points related to equilibrium and to time t = 0 min (i.e., using
10 < t < 60 min), we can obtain a very good fit to the data, and the model of first order becomes
log (7.5 - qt) = 0.564 - 0.020 t, with a residue parameter R2 = 0.9981. Thus, this model is
acceptable to describe the kinetics of ASA adsorption by the sample Si-MCM-41. Figure 11
presents the fits for this case.

INSERT FIGURE 11

For the kinetic model of second order, equation (1) becomes equation (4):

= 2 . ( )2

(4)

A linearization is also possible, and is represented by equation (5):

=
+
. 2

(5)

Thus, Table 7 was built in order to obtain the value of k2 and replace it in the model. The
value of qe was maintained as previously done. The procedure, as applied to experimental data
gave the results shown in Figure 12.

INSERT TABLE 7

INSERT FIGURE 12

The values obtained for the residue (R2 = 0.99988), very close to the unit, indicated an
excellent fit of the data obtained by the second order model. Thus, it can be concluded that the
adsorption kinetics of ASA on the Si-MCM-41 is a kinetic process which is well described by
the model (t / qt) = 4.318 + (t / 7.5). Thus, the kinetic model of second order is slightly more
suited to data that the model of order 1. More experiments to confirm this result would be
needed. If the model of 2nd order is accepted, this suggests that the step limiting of velocity is
the adsorption on the adsorbent surface, and points to the chemical nature of the adsorbateadsorbent interactions.

3.3.3. Effective diffusion coefficient for the AAS / Si-MCM-41 system

To calculate the effective diffusion coefficient of the AAS in the molecular sieve
Si-MCM-41 (sample A3), we followed the procedures proposed by Valencia (2013), who
performed the adsorption of vitamin B12 on a metal-organic mesoporous system (MOF).
Initially a mathematical adjustment was made for the loading curve of ASA molecular
sieve (Figure 10), yielding the following equation:

 = 4,175. (

3,200 )

3,286.

( 21,664)

+ 7,461
(6)

which residue was R2 = 0.9994, x is the loading time for ASA and F(x) the relative mass
adsorbed. Using equation (6) it is possible to determine the initial loading rate and the
instantaneous rate for any time. The shape of the curve suggests that the initial loading rate is
constant until about 15 min. We can, in this way, from the differentiation of this equation with
respect to time to calculate this rate (F(x)). Then

() = 1,305.

3,200

+ 0,152.

( 21,664)

(7)
To determine the initial loading rate we use the point (t=10 min, 5.2% mgASA/mgsolid):
F(10 min) = 0,153 % mgASA/(mgsolid.min)
According to Chowdhury et al. (2005), the diffusion coefficient of a molecule diffusing
through a mesopores structure can be found from measurements of solution concentration as a
function of time. In this study we used a spherical geometry for the particle for the resolution of
the model, which is an approximation, because the particles have irregular shapes as seen in the
micrographs of Figures 4a and b.
An approximation of the true value of the diffusion coefficient for this system can be
calculated from the following equation (Bird et al. 2006):

=
+
+1

6
2 2
=1 9 1 + +

(8)

In this model, the equation has been cast in dimensionless form through the use of
Laplace transforms. The variables in this equation are defined as follows:

, =
, =
0

(9)

where Ct is the ASA concentration in ethanol solution at time t, C0 is the initial ASA
concentration in the solution, Vf is the volume of the solution, Vb is the average volume of the
particle size of Si-MCM-41, Deff is the effective diffusion coefficient and R is the radius of the

particle. It was assumed in this case that the partition coefficient () equals 1. In this model i
are the solutions of equation (10):
tan =

3
3 + 2

(10)

Once is determined, can be found as the zeros of the function

= tan

3
3 + 2

(11)

which is the Laplace transform of equation (10). From equations (8) and (9) we have:

+ 1

(2
)
2
)

1+
+ ( )2 . 2

6(

=1

(12)

And when

and the system reaches the equilibrium and the infinite sum in the second term on the
right side of equation (12) tends to zero and then / ( +1). It was assumed the
concentration after 240 min of loading as the equilibrium concentration. It provides:
92,5 / 100 = / ( +1)

and

= 12,333

(13)

Using this value for , the i values were calculated from equation (9). Figure 13 shows
the graph of s versus G(s) and Table 8 presents the 7 first approximations for i with G(s) = 0.

INSERT FIGURE 13

INSERT TABLE 8

With the values of i and , we can calculate the value of Deff from equation (12).
Substituting the known values and using the data collected for the initial loading step, we have:

94,8 12,333
=
+
100 13,333

0,948 =

12,333
+
13,333

0,0230 =
=1

=1

=1

6(12,333)

(2 10 . )
50 2

9 1 + 12,333 + 12,3332 . 2

(14)
(2 10 . )
50 2
6(12,333)

9 1 + 12,333 + 12,3332 . 2

(15)

73,998 . (0,004. )
119,997 + 152,103. 2

(16)

Using the equation (16), the value of the effective diffusion coefficient, Deff, was obtained
from successive approximations and interpolations in the spreadsheet.
Finally we have for ASA/Si-MCM-41 system under the conditions of this work:
Deff = 5,02 x 10-17 m2/s
This value is consistent with the data found by Polakovic et al. (2005) who performed the
adsorption of aspirin on activated carbon at 30 oC, finding a value of 0.8 x 10-17 m2/ s. Pendleton
and Park (2011) when adsorbing allyl isothiocyanate in SBA-15 found a value for Deff between
0.3 x 10-17 and 0.44 x 10-17 m2/s. As we have different materials in different conditions, this data
is acceptable.

4. CONCLUSION

All the molecular sieves employed in this study are suitable for the adsorption of ASA.
Among the zeolites, HUS-zeolite Y adsorbs higher ASA due to their larger specific area and
volume, but in relation to the available area, was less efficient. Silicalite-1 showed to be less
efficient for this process possibly due to absence of aluminum and loading sites.
Among the Si-MCM-41samples, An and Tn, the samples synthesized with Aerosil (An) were most
effective for the adsorption of ASA, probably due to the higher number of silanol groups that this kind of
silica source generates in these samples, which seems to be related with a better adsorption of ASA.

Thermogravimetric analyses proved that the HUS-Y zeolite adsorbs more ASA among
the zeolites and A3 and T1 samples of Si-MCM-41 were the better for this function. The
behavior of the samples An and Tn demonstrates that the interaction between the ASA and the
Si surface MCM-41 is not directly linked to the specific area or volume available for adsorption,
but by the amount of adsorption sites that have important role in the adsorptive behaviour of the
sample. Probably the surface silanol groups more abundant in samples An, are the key to
explain the improved performance of these samples.
The solvent is adsorbed in large proportion in relation to the ASA. A study with different
solvents is necessary to determine what would be the best option, because the interaction of the
solvent with ASA competes in adsorption. The formation of a possible clathrate ASA-Ethanol
positively charged could explain the better affinity with the samples with aluminum in the
lattice.
Regarding the kinetics of adsorption of ASA, it was observed that the loading reaches its
peak only after 15 min of contact with Si-MCM-41 (sample A3). The methodology employed
data obtention was very appropriate and the experiment was considered valid. The adsorption
kinetics using a first or second order models fits well to the proposed experiment, but second
order model is the better, providing the following equation: (t/qt) = 4.318 + (t/7.5), with an
residue parameter of R2 = 0.99988.
In the experiment for determining the effective diffusion coefficient for the
ASA/Si-MCM-41, the initial rate of loading (0.153% mgASA/mgsolid.min), the equation for the
rate of adsorption (eq. 7) and the effective diffusion coefficient (Deff = 5.02 x 10-17 m2/s) were
obtained. This indicates that the Si-MCM-41 is a porous material that has great capacity for
ASA adsorption and low hindering to diffusion of the drug.
Whereas the model used for the calculation of the effective diffusion coefficient is only
an approximation, a number of recommendations can be made to provide a more accurate result,
since it was adopted a spherical geometry to particles and a perfect interaction of the solution
ASA-ethanol. The first aspect can be improved by using data from particle laser analyser to
more accurately measure the particle sizes and their distribution. The adsorption experiment

could be carried out under the action of an ultrasound system, which could be coupled in a
circulating system with a sampling point to prevent the agglomeration of particles and formation
of dead zones.

5. ACKNOWLEDGMENTS
We would like to thank the Catalysis Laboratory (LabCat-UFSCar) and

Electronic

Microscope Laboratory (LME-UFSCar), both located in the Federal University of So Carlos

(Brazil), for the XRD measurements and for SEM and TEM analyses.

6. REFERENCES
Amos, H.E., Wilson, D.V., Taussig, M.J. and Carlton, S.J. (1971) Clin. Exp. Immunol. 8, 563.
Beninati, S., Semeraro, D., Mastragostino, D. (2008) Adsorp. Sci. Tech. 3 (9), 721.
Beck, J.S., Vartuli, J.C., Roth, W.J., Leonowicz, M.E., Kresge, C.T., Schmitt, K.D., Chu,
C.T.W., Olson, D.H., Sheppard, E.W., McCullen, S.B., Higgins, J.B. and Schlenker, J.L.
(1992) J. Am. Chem. Soc. 114 (27), 10834.
Bird, R.B., Stewart, W.E., Lightfoot, E.N. (2006) Transport Phenomena, John Wiley & Sons,
New York / London.
Blin, J.L., SU, B.-L. (2002) Langmuir 18 (13), 5303.
Brunel, D., Cauvel, A., Di Renzo, F., Fajula, F., Fubini, B., Onida, B., Garrone, E. (2000) New
J. Chem., 24 (10), 807.
Brinker, C.J. (1996) Current Opin. Sol. State & Mater. Sci. 1, 798.
Cai, Q., Lin, W.-Y., Xiao, F.-S., Pang, W.-Q., Chen, X.-H., Zou, B.-S. (1999) Microp. Mesop.
Mater. 32, 1.
Carvalho, W.A. (1997) Oxidao seletiva de hidrocarbonetos saturados catalisada por metais
de transio incorporados em peneiras moleculares do tipo MCM-41. PhD Thesis,
Universidade Estadual de Campinas, Campinas / Brazil.
Charnay, C., Begu, S., Tourn-Pteilh, C., Nicole, L., Lerner, D.A., Derrien, A., Renard, G.,
Brunel, D. (1998) Stud. Surf. Sci. Catal. 117, 445.
Charnay, C., Begu, S Tourn-Pteilh, C., Nicole, L., Lerner, D.A., Devoissele, J.M. (2004) Eur.
J. of Pharm. and Biopharm. 57, 533.
Datt, A. (2012) Applications of mesoporous silica and zeolites for drug delivery. PhD Thesis,
University of Iowa, Iowa / EUA.
Cheng, C.-F.; Ho, D.; Klinowski, J. (1997) J. Chem. Soc. Faraday Trans. 93(1), 193.
Chowdhury, M., Hill, D., Whitaker, A. (2005) Aust. J. Chem. 451.
Gerhardt, C. (1975) Manufacturing from salicylic acid and acetic anhydride: Faith, Keyes and
Clarks. Industrial Chemicals, New York/Lowenhein.
Giannetto, G. (1990) Zeolitas: Caracteristicas, propiedades y aplicaciones industriales,
Ediciones Innovacin Tecnolgica, Caracas/Venezuela.
Gonzalez, G., Sagarzazu, A. and Zoltan, T. (2013) J. of Drug Deliv. 2013, 1.
Horcajada, P., Ramila, A., Perez-Pariente, J., Vallet-Regi, M. (2004) Mic. Mesop. Mat. 68, 105.
Grunwald, E. and Berkowits, B.J. (1951) J. Am. Chem. Soc. 73, 4939.
Meynen, V., Cool, P., Vansant, E.F. (2009) Microp. Mesop. Mater. 125, 170.
Moller, K. and Bein, T. (1998) Chem. Mater. 10, 2950.
Mollo, L., Levresse, V., Ottaviani, M.F., Ellouk-Achard, S., Jaurand, M.C., Fubini, B. (1997)
Environ. Health Perspect. 105, 1031.
Nema, S., Washkuhn, S., Brendel, R.J. (1997) J. Pharm. Sci. Technol. 51, 166.
Prakash, A.M. and Kevan, L. (1998) J. Am. Chem Soc. 120, 13148.
Rakic, V., Rajic, N., Dakovic, A., Auroux, A. (2013) Microp. Mesop. Mater. 166, 185.
Unger, K., Rupprecht, H., Valentin, B., Kircher, W. (1983) Drug Dev. Ind. Pharm. 9, 69.
Uglea, C.V., Albu, I., Vatajanu, A., Croitoru, M., Antoniu, S., Panaitescu, L., Ottenbrite, R.M.
(1994) J. Biomater. Sci. Polym. 6, 633.
Valencia, V. (2013) Und. J. Math. Model. 5 (1), 3.
Vallet-Regi, M., Ramila, A., del Real, R.P., Prez-Pariente, J. (2001) Chem. Mater. 13, 308.
Watson, H.R., Berquist, D. (2000) Eur. J. Vasc. Endovasc. Surg. 19, 445.
Zhou, C.F. and Zhu, J.H. (2005) Chemosphere 58, 109.

LIST OF FIGURES

Figure 1. Molecule of acetylsalicylic acid and its dimensions calculated by the method MM2
(adapted from Beninati et al., 2008).

1000 cps
Sample T3

10

1000 cps
Sample T2

10

1000 cps
Sample T1
0

10

2/degrees

Figure 2. Diffraction patterns of calcined samples of Si-MCM-41 synthesized with TEOS at

100 C for 2 days.

1000 cps

Sample A3
8

10

20000 cps
Sample A2
0

10

4000 cps
Sample A1
0

10

2/degrees

Figure 3. Diffraction patterns of calcined samples of Si-MCM-41 synthesized with Aerosil,

T = 150 C (thermal treatment (days): A1 = 2 , A2 = 4, A3 = 10).
(a)

(b)

Figure 4. (a) Si-MCM-41 (Sample A2) synthesized with Aerosil. (b) Si-MCM-41 (sample T2)
synthesized with TEOS.

Figure 5. Micrograph (TEM) of sample A2 synthesized with Aerosil, view along [010] axis.

200

150

Adsorbed Volume (cm /g) - STP

ZSM-5
ZSM-5+ASA

100

50

0
0,0

0,2

0,4

0,6

0,8

1,0

P/Po

Figure 6. Physisorption isotherms of HZSM-5 before and after loading (24 h).

600

400

Adsorbed Volume (cm /g) - STP

500

A2

300

A2 + ASA

200
100
0
0,0

0,2

0,4

0,6

0,8

1,0

P/Po

Figure 7. Physisorption isotherms of sample A2 before and after loading (24 h).

600

400

Adsorbed Volume (cm /g) - STP

500

T2

300

T2+ASA

200
100
0
0,0

0,2

0,4

0,6

0,8

1,0

P/Po

Figure 8. Physisorption isotherms of sample T2 before and after loading (24 h).

A3

100

Relative Mass Loss (%)

A2
A1

95

90
100

200

300

400

500

600

700

Temperature ( C)

Figure 9. Thermogravimetric curves for samples An (Si-MCM-41, Aerosil).

Adsorbed Mass ASA (%)

8
7
6
5
4
3
2
1
0
0

50

100

150

200

250

Time (min)

Figure 10. ASA adsorption curve as a function of time - adsorption kinetics.

Equation

1,0

y = a + b*

Adj. R-Square

0,99818
Value

log(qe-qt)

0,5

Intercept

Slope

Standard Error

0,5648

0,01794

-0,02097

5,16667E-4

0,0

-0,5

-1,0
0

20

40

60

80

100

120

140

160

180

200

t(min)

Figure 11. First order kinetic model of ASA adsorption applied to the experimental data:.
Transient adjustment range.

35

Equation

y = a + b*x

Adj. R-Square

0,99988

30

t / qe

25

Value
B

Intercept

Slope

Standard Error

-0,57342

0,084

1,01597

0,00497

20
15
10
5
0
0

10

15

20

25

30

35

t / qt

Figure 12. Second order kinetic model of ASA adsorption applied to the experimental data.

Figure 13. Graph of G(s) as a function of s defined by equation (11).

LIST OF TABLES
Table 1 Data from microporous molecular sieves employed in this work.
Cat.

SiO2/Al2O3

Vm (cm3/g)

S (m2/g)

MSA (m2/g)

HZSM-5
HUS-Y

26,3
5,62

0,134
0,279

385
664

91
65

Vm = specific micropore volume; S = specific surface area; MSA = mesoporous surface area.

Table 2. Properties of the samples of Si-MCM-41 from their respective diffraction patterns.
Sample

Go (%)

FWHM

G0

2100

2110

2200

aH ()

T1

69

0,50

136

2,30

44,4

T2

61

0,18

336

2,46

4,32

5,00

41,4

T3

62

0,20

311

2,27

3,93

4,52

45,0

A1

37

1,90

53,6

A2

100

0,50

201

2,63

4,44

5,19

38,8

A3

100

0,93

108

1,55

65,8

Go = organization degree = (I100/I100 pattern)x100; FWHM = full width at half maximum (plane (100));
Go = Go/FWHM; 2100 = position 2 (plane (100)); 2110 = position 2 (plane (110)); 2200 = position 2 (plane
(200)); aH = hexagonal framework parameter.

TABLE 3. Textural parameters of the samples before and after loading with ASA.
S(m2/g)

Vo(cm3/g)

V(cm3/g)

323

47

146

143

146

134

114

70

593

387

206

236

158

T1

1478

1387

91

617

T2

1340

1219

121

470

T3

1147

1030

117

511

Sample

So(m2/g)

S(m2/g)

Silicalite-1

370

HZSM-5

280

HUS-Y

V(cm3/g)

V(%)

DP()

2,1

~ 5,6

44

38,6

~ 5,6

78

33,1

~7,4

576

41

6,6

33,5

428

42

8,9

37,2

446

65

12,7

37,8

A1

23

26

50

35

15

30,0

37,1

A2

1400

1162

238

541

448

93

17,2

37,6

A3

571

191

380

329

128

201

61,1

33,4

So = specific surface area before loading; S = specific surface area after loading; V o = specific volume total of pores
before loading; V = specific volume total of pores after loading; S = variation in specific surface area (S-So);
V = variation in specific volume total of pores (V-Vo); V (%) = 100 x V / Vo; DP = diameter of pores of
molecular sieve.

Table 4. Percentage of adsorbed ASA and solvent in the molecular sieves from TG analyses.
Molecular sieve
Zeolites

Si-MCM-41

Sample
Silicalite 1
HZSM 5
HUS Y
A1
A2
A3
T1
T2
T3

ASA (%)
2,26
5,87
11,71
4,10
4,48
4,73
5,24
3,11
4,51

Solvent (%)
0,67
4,68
18,19
4,10
2,61
4,74
5,74
6,19
3,83

Total Mass Loss (%)

2,93
10,55
29,90
8,20
7,09
9,47
10,98
9,30
8,34

Table 5. Data obtained by thermogravimetry for the adsorption kinetics experiment.

Time (min)
0

m0 (mg)
-

m120 (mg)
-

mf (mg)
-

mAAS (mg)
-

mAASrel (%)
0

10

1,454

1,418

1,348

0,071

5,2

15

0,843

0,834

0,788

0,046

5,8

30

0,715

0,698

0,655

0,043

6,6

60

0,802

0,821

0,765

0,056

7,3

180

0,858

0,840

0,782

0,058

7,4

240

2,181

2,205

2,058

0,153

7,5

m0 = initial mass of the loaded sample; m120 = m0-msolvent; mf = final mass; mASA = mass of ASA;
mASArel = mass relative of ASA = mASA / mf.

Table 6. Parameters obtained from the kinetic model of adsorption of first order.

t (min)
0
10
15
30
60
180
240

log(qe-q)
0,875
0,362
0,230
-0,046
-0,699
-1
-

k1 (min-1)
0,118
0,099
0,071
0,060
0,024
0,008
k1m = 0,047

Table 7. Parameters obtained from the kinetic model of adsorption of second order.
t (min)

t/qt
(min.mgsolid/ mgAAS)

0
10
15
30
60
180
240

t/qe
(min.mgsolid/ mgAAS)

1,923
2,586
4,545
8,219
24,324
32,000

k2
(mgAAS/mgsolid)

0
1,333
2,000
4,000
8,000
24,000
32,000

0,074
0,054
0,030
0,016
0,006
0,004
k2m = 0,031

Table 8. i values determined from equation (11) for G (s) = 0.

i

0,50

4,38

7,53

10,50

13,50

20,00

23,00