In vitro adsorption study of fluoxetine in activated carbons and activated carbon fibres

F U E L P R O CE SS I NG T EC H NOL O G Y 8 9 (2 0 0 8 ) 5 4 9–5 5 5

w w w. e l s e v i e r. c o m / l o c a t e / f u p r o c

In vitro adsorption study of fluoxetine in activated carbons and activated carbon fibres J.M. Valente Nabais ⁎, A. Mouquinho, C. Galacho, P.J.M. Carrott, M.M.L. Ribeiro Carrott Centro de Química de Évora e Departamento de Química da Universidade de Évora, Rua Romão Ramalho no. 59, 7000-671 Évora, Portugal

AR TIC LE I N FO

ABS TR ACT

Article history:

We study the in vitro adsorption of fluoxetine hydrochloride by different adsorbents in

Received 6 July 2007

simulated gastric and intestinal fluid, pH 1.2 and 7.5, respectively. The tested materials were

Received in revised form

two commercial activated carbons, carbomix and maxsorb MSC30, one activated carbon

26 October 2007

fibre produced in our laboratory and also three MCM-41 samples, also produced by us.

Accepted 29 October 2007

Selected samples were modified by liquid phase oxidation and thermal treatment in order to change the surface chemistry without significant modifications to the porous

Keywords:

characteristics. The fluoxetine adsorption follows the Langmuir model. The calculated Q0

Activated carbons

values range from 54 to 1112 mg/g. A different adsorption mechanism was found for the

Fluoxetine

adsorption of fluoxetine in activated carbon fibres and activated carbons. In the first case

Activated carbon fibres

the most relevant factors are the molecular sieving effect and the dispersive interactions

Bioadsorption

whereas in the activated carbons the mechanism seams to be based on the electrostatic

Liquid phase adsorption

interactions between the fluoxetine molecules and the charged carbon surface. Despite the different behaviours most of the materials tested have potential for treating potential fluoxetine intoxications. © 2007 Elsevier B.V. All rights reserved.

1.

Introduction

Modern society and the stressing daily life in a large part of the world lead to a significant prevalence of depression, which, for example in the USA can reach 10% in the elderly [1]. According to the latest report on depression from the World Health Organization (WHO) published in 2001 an estimated 340 million people around the world suffer from the disease [2]. One way of treating the condition is with drugs. Several types of antidepressant drugs are available, and in many countries they are among the most commonly prescribed medicines. One family of antidepressants, called selective serotonin reuptake inhibitors (SSRIs), was introduced in the late 1980s. The name of these drugs comes from their effect, which is to prevent the removal (reuptake) from the nerve

endings of one type of chemical (serotonin) that is important for transmitting nerve impulses between brain cells. Among SSRIs fluoxetine is one of the most widely used. Fluoxetine was developed by the pharmaceutical firm Eli Lilly under the commercial brand name Prozac®, which was approved by the Food and Drug Administration in 1987 for the treatment of depression (including paediatric depression), obsessive–compulsive disorder (in both adult and paediatric populations), bulimia nervosa, panic disorder and premenstrual dysphonic disorder. The recommended dose ranges from 20 to 80 mg daily. The molecule normally used in pharmaceutical formulations is fluoxetine hydrochloride, IUPAC name N-methyl-3phenyl-3-[4-(trifluoromethyl)phenoxy]propan-1-amine hydrochloride, shown in Fig. 1. Although fluoxetine hydrochloride (fluoxetine HCl) is most widely marketed as Prozac® about 30 countries and at least 48

⁎ Corresponding author. Tel.: +351 266745318; fax: +351 266745303. E-mail address: [email protected] (J.M.V. Nabais). 0378-3820/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.fuproc.2007.10.008

550

F U E L P RO CE SS I NG T EC H NOL O G Y 8 9 (2 0 0 8) 5 49 –5 5 5

Fig. 1 – Fluoxetine hydrochloride.

manufacturers and countless pharmacies have also marketed several medicines using the molecule [3], since Eli Lilly's patent on Prozac® expired in 2001. Milane et al. [1], have analyzed the associations between suicide rates and dispensing of the prototypic SSRI antidepressant fluoxetine in the United States during the period 1960–2002. The authors found that the introduction of SSRIs was temporally associated with a substantial reduction in the number of suicides. On the other hand, it is reported that a high suicide attempt rate of depressed people is executed by overdose of the medicine used to control the condition [4]. For instance, in Portugal this type of suicide increased from 68.7% in 2000 to 71.5% in 2005. Additionally, accidental overdoses are much more frequent having grown from 28.1% in 2000 to about 40.7% in 2004. Also, according to the Health Minister, in Portugal, 8 people are admitted every day to hospitals victims of medicinal intoxication due to self-medication or suicide attempts and about 70% of these hospitalisations are related to abusive use of antidepressive drugs. There is no antidote for fluoxetine overdoses and thus the treatment of choice for acute drug intoxication is the use of activated carbon by oral administration or instillation by nasogastric tube of an aqueous suspension of activated carbon, which will adsorb the drug at both gastric and intestinal tracts preventing the drug from entering the blood circulation and thus the systemic toxicity. Sometimes after the activated carbon administration a gastric lavage is performed [5]. The drug adsorption onto the activated carbon depends on several factors such as pH, temperature, adsorbent acidity, surface area, pore volume, mean pore width and particle size, drug solubility and ionisation and other drugs and substances present in gastrointestinal tract. Carbon materials have great potential to be used in medicinal applications due to properties such as good biocompatibility, chemical inertness, nontoxicity and no immune reaction with the human body. Although all the facts exposed are documented, the publication of papers about the adsorption of fluoxetine in carbon materials is very rare. Filling this gap was the main motivation to perform our study and the main objective was to increase our knowledge about the mechanisms involved in fluoxetine adsorption.

2.

Experimental

2.1.

Materials

Two commercial activated carbons were used. One sample, Carbomix (carb) from Norit N.V. the Netherlands, is commercially available for use in medicinal applications. The other sample, Maxsorb MSC30, is from Kansai Coke & Chemicals Co, and is made from petroleum coke and chemically activated with KOH. The activated carbon fibre (ACF), F953, was prepared by us from a commercial acrylic textile fibre, provided by Fisipe (Barreiro, Portugal). For the production of the ACF about 12 g of fibre and a horizontal tubular furnace were used. Stabilisation of the fibres was carried out by heating to 300 °C at a rate of 1 °C min− 1 under a constant N2 flow of 85 cm3 min− 1 and maintaining for 2 h. The fibre was then carbonised by raising the temperature at a rate of 5 °Cmin- 1 to 800 °C and maintaining at that temperature for 1 h. Activation was carried out by raising the temperature again by 15 °Cmin- 1 to 900 °C and then switching to a CO2 flow of 85 cm3 min− 1, maintaining for 5 h in order to obtain 53% burn-off, switching back to the N2 flow and allowing to cool to below 50 °C before removing the ACF from the furnace and storing in a sealed sample flask (details given elsewhere, Carrott et al. [6]. The materials were treated in order to change the surface chemical properties to obtain a more acidic and basic carbon, respectively by using liquid phase oxidation with concentrated nitric acid and thermal treatment at 900 °C. The sample designations of the treated sample are: carb ox, F953 ox and carb red, respectively. The adsorption of fluoxetine onto ordered mesoporous materials was also studied. The Ti–MCM-41 samples were prepared by direct synthesis at ambient temperature and pressure, using tetraethoxysilane, titanium n-butoxide in ethanol or titanium ethoxide in propan-2-ol, hexa-or octadecyltrimethylammonium bromide and ammonia. The synthesis procedure was previously reported [7,8]. The Ti–MCM-41 samples are designated by TiRan-x, where R = B or E denotes the metal alcoxide, (Ti(OR)4), respectively, titanium n-butoxide and ethoxide, a = e or p represents the alcohol, respectively, ethanol (EtOH) or propan-2-ol (2-PrOH), n = 16 or 18 refers, respectively, to hexa-or octa-decyltrimethylammonium bromide (C16TMABr or C18TMABr) used as structure-direct agents and x corresponds to the nominal molar ratio Si/Ti. The Al–MCM-41 sample, designated by Al–MCM-41(30), was also prepared by direct synthesis at ambient temperature and pressure, using tetraethoxysilane, aluminium sulphate, hexadecyltrimethylammonium bromide and ammonia with an Si/Al = 30. The synthesis procedure was previously described [9]. In the work now reported we used the samples TiEp18-5, TiBe16-50 and Al–MCM-41 (30).

2.2.

Materials characterisation

Nitrogen adsorption isotherms at −196 °C were determined using a CE Instruments Sorptomatic 1990 after outgassing the samples at 400 °C to a residual vacuum of 5 × 10− 6 mbar. The concentration of surface acid and basic sites on the materials,

551

F U E L P R O CE SS I NG T EC H NOL O G Y 8 9 (2 0 0 8 ) 5 4 9–5 5 5

spectrophotometer. The absorbance of solutions up to 80 mg/L obeys Beer's Law. Linear regression indicated linearity with correlation coefficient, r2, of 0.999. The adsorption kinetics was evaluated by preparing suspensions of adsorbent (0.1 g) with 20 mL of fluoxetine stock solution. The suspensions were placed in a shaking thermostat bath at 37 °C and the residual fluoxetine concentration determined at prescribed time intervals between 20 and 240 min. Batch adsorption experiments were carried out in a series of Erlenmeyer flasks of 100 mL capacity covered with Teflon sheets inserted in a shaking thermostat bath at 37 °C for 120 min using 0.1 g of adsorbent and 20 mL of fluoxetine solutions with variable concentrations, usually between 0.1 and 1 g/L, and at both pH 1.2 and 7.5. In some studies higher concentrations, up to 3 g/L, were used. In these cases two more stock solutions of 3 g/L were prepared according to the procedure already described.

Fig. 2 – Representative nitrogen adsorption/desorption isotherms at −196 °C.

3.

Results and discussion

3.1.

Materials characterisation

CHA and CB, were determined by equilibration during 24 h with 0.01 M NaOH or 0.01 M HNO3 (0.15 g of fibre for 30 mL of solution) and determination of the excess by back titration. The point of zero charge was determined by mass titrations, details given elsewhere [6].

All carbon samples used as received or as prepared (carbomix, maxsorb and F953) have basic properties with pzc values between 8.01 and 9.50 and a wide range of porous characteristics. In order to expand our knowledge about the adsorption mechanism more relevant in the fluoxetine adsorption these materials were modified by changing the surface chemistry without any significant modifications to the porous structure, as shown later. The nitrogen adsorption isotherms for the carbon materials, shown in Fig. 2, are all Type I according to the IUPAC classification [10]. Although, for reasons of simplicity, we only show the isotherms of the materials prior to the treatments, we can say that all the isotherms of the treated samples have the same profile as the samples before the treatments. All the isotherms obtained, are reversible and have very low slope in the multilayer region indicating a low external surface area and the absence of significant mesoporosity, as also indicated by the calculated values for Vs and V0, which are very similar. The results of isotherm analysis by the Brunauer–Emmett– Teller (BET), αs (using standard data published by Carrott et al. [11] and Dubinin–Radushkevich (DR) methods are shown in Table 1. According to the relative pressure range over which

2.3.

Fluoxetine adsorption study

Fluoxetine hydrochloride (reference standard, pKa 9.5, 345.79g mol− 1) was kindly provided by Eli Lilly. The adsorption study was performed at two simulated real situations, namely gastric and intestinal fluids, with pH 1.2 and 7.5, respectively. A stock solution of fluoxetine HCl (1 g/L) was prepared in simulated gastric fluid (pH 1.2), pepsin omitted, (SGF) which consists of 2,0 g/L NaCl and 7 mL/L of concentrated HCl. Other stock solution of 1 g/L was prepared in simulated intestinal fluid (pH 7.5), pepsin omitted, (SIF) which consists of 50 mL NaH2PO4 0.1 M and 42.5 mL NaOH 0.1 M for 1 L of solution. Standard solutions were obtained by diluting the stock solutions with the simulated fluids, in order to maintain the solution pH. The determination of fluoxetine HCl was done by ultraviolet absorption at 226 nm using a Nicolet evolution 300 Uv–Vis

Table 1 – Textural and chemical characterisation of the carbon materials Sample

Porosity 2

−1

αS

ABET/m g

Carb Carb ox Carb red F953 F953 ox MSC30

1396 1506 1314 791 811 2487

pzc DR

Vs/cm3 g− 1

Aext/m2 g− 1

V0/cm3 g− 1

L0/nm

0.70 0.75 0.64 0.41 0.41 1.09

71 70 53 7 12 40

0.52 0.56 0.46 0.36 0.36 0.97

1.53 1.29 1.26 0.91 0.96 1.74

8.01 3.70 9.52 9.50 6.09 8.30

Group concentration/ mmol g− 1 Acidic

Basic

0.85 3.82 0.46 1.26 1.12 0.34

0.61 0.07 0.03 0.63 0.01 0.19

552

F U E L P RO CE SS I NG T EC H NOL O G Y 8 9 (2 0 0 8) 5 49 –5 5 5

the equations are applied we can consider that the αS method gives an evaluation of the total pore volume and the DR method provides an evaluation of the primary micropore volume. Primary micropores can be defined as the pores with mean pore width less than 0.8 nm in which the physisorption is associated with enhanced adsorbent-adsorbate interactions that take place at very low p/p0 in pores with molecular dimensions (less than approximately two times the molecular dimension of the adsorbate, usually nitrogen, as in our case). Secondary micropores have mean pore width between 0.8 and 2 nm and the pore filling occur by a cooperative adsorption mechanism at somewhat higher relative pressures [10,12]. Therefore, the difference between Vs and V0 can be considered an estimate of the secondary micropore volume. The isotherms of the ordered mesoporous materials used in the study are also shown in Fig. 2. According to the IUPAC classification [10] both samples present type IVc nitrogen adsorption isotherms typical of MCM-41 materials exhibiting pore filling steps which occur, in each case, within a fairly narrow range of p/po indicating size uniformity of the tubular unidirectional mesopores. Considering the results of the TiBe16-50 and Al–MCM-41(30) sample it can be seen that for high relative pressures, after pore filling was complete, the isotherm is almost horizontal and completely reversible suggesting reasonable uniform particle morphology and the absence of significant interparticle agglomeration. For higher Ti content sample, TiEp18-5, we can notice an increase in external surface area and the presence of secondary mesoporosity due to interparticle agglomeration. The isotherm still presents a vertical step, indicating reasonable pore size uniformity of the mesostructured part, but this must exist in a more reduced amount in comparison to the other sample as the step is shorter. By XRD patterns it was possible to observe that the materials possess a high degree of ordering of the pore structure and well-formed hexagonal pore arrays [7,8,13]. The ACF sample F953 is the material with the smallest BET apparent surface area (ABET), pore volume and with the narrowest porosity, as indicated by the very similar Vs and V0 values shown in Table 2. Although the porous structure of this sample is less developed it has the advantage of having a

different porous network, when compared with the activated carbon samples, with the micropores directly accessible to the adsorbing molecules [6] which give to ACFs the advantage of having, in most cases, high adsorption/desorption rates. At the other extreme is the maxsorb (MSC30) sample with an extremely high value of BET apparent surface area and significant pore volume but also with a reduced pore size distribution. The carbomix carbon has a wide range of pores that goes from primary micropores to small mesopores. Additionally to this broad porosity carbomix also has a relatively high value of surface area. The calculated values of total surface area (As), mesopore volume (Vp) and pore width for samples TiBe16-50 and TiEp18-5 are respectively, 1003 and 807 m2/g, 0.77 and 0.62 cm3/g and 3.8 and 4.3 nm. For the aluminium substituted sample the corresponding values are 986 m2/g, 0.72 cm3/g and 3.6 nm. The oxidation by nitric acid in liquid phase caused a significant impact on the chemical properties of the carbomix and F953 samples by producing more acidic samples, as can be seen in Table 1. The pzc value decreases from 8.01 to 3.70 in the first case and from 9.50 to 6.09 in the ACF sample. On the other hand, the thermal treatment of carbomix leads to an increase of the pzc value to 9.52. Both treatments originate modifications in the surface group concentration. In general terms, the oxidation enhances the concentration of the functional groups with acidic properties, which are able to neutralize NaOH, and reduces the groups with basic characteristics, neutralized by HNO3. The thermal treatment leads to the release of most of the heteroatoms present on the material surface and thus to a diminution of their concentration. During the cooling the material suffers a partial reoxidation of the surface with the creation of basic type functional groups such as pyrone groups [14]. The samples textural properties did not show any significant alteration after the treatments undertaken. The data in Table 1 shows that the values of BET apparent surface area, pore volume and type of porosity did not substantially change. For instance, we can observe an alteration of about 7% in BET apparent surface area.

3.2. Table 2 – Fluoxetine adsorption data Sample pH Nadsmax/ FC ACC mg g− 1

Carb

1.2 7.5 Carb ox 1.2 7.5 Carb red 1.2 7.5 F953 1.2 F953ox 1.2 7.5 max 1.2 7.5

408 626 542 727 398 468 100 105 54 826 1095

+ + + + + + + + + + +

+ (0)+ + − ++ + ++ + − ++ (0)+

Langmuir b/L mg− 1

Q0/mg g− 1

r2

0.083 0.092 0.145 0.189 0.096 0.304 0.023 0.042 0.059 0.184 0.375

416 625 556 714 400 476 103 97 55 833 1112

0.999 0.999 0.999 0.999 0.999 0.999 0.989 0.995 0.998 0.999 0.999

FC — fluoxetine charge ACC — materials mean charge (++ highly positive, − negative, + positive, (0)+ neutral or slightly positive).

Fluoxetine adsorption study

The fluoxetine adsorption rate for carbomix and F953 is very high, as can be seen in Fig. 3, which is fundamental for the treatment of intoxications, because to be effective the adsorbent must have a high adsorption rate in order to prevent the absorption of the drug by the human body. For carbomix at both pH values and for F953 at pH 7.5 the equilibrium is reached in about 10 min. For F953 at pH 1.2 it takes a longer time until the equilibrium is reached, about 100 min. Therefore, in the batch adsorption trials we used 120 min as stirring time to ensure that a complete equilibrium is reached in all trials. The fluoxetine adsorption study was performed at normal body temperature, 37 °C, and in both simulated gastric and intestinal fluids because these are the places where the fluoxetine is absorbed by the human organism. The adsorption isotherms at 37 °C shown in Fig. 4 can be considered, according to the Giles classification [15], of the Langmuir type with the formation of a well defined plateau at

F U E L P R O CE SS I NG T EC H NOL O G Y 8 9 (2 0 0 8 ) 5 4 9–5 5 5

Fig. 3 – Adsorption kinetics curves for carbomix and F953.

moderate fluoxetine concentration values. The results were analysed by the Langmuir model using the linear form of the model Eq. (1). The characteristic constants shown in Table 2 were calculated from the best-fit lines to the experimental data. Ce =qe ¼ ð1=ðQ0 bÞÞ þ ð1=Q0 ÞCe

ð1Þ

where Ce is the equilibrium concentration (mg/L), qe is the adsorbed quantity (mg/g); Q0 (mg/g) and b (L/mg) are characteristic constants. Typically, b is related to the enthalpy and intensity of adsorption and Q0 is associated with the maximum adsorption capacity [25]. The Langmuir model fits the experimental data very well, as can be seen in Table 2, with excellent correlation coef-

553

ficient, r2, always superior to 0.99, and with the calculated Q0 value very similar to the maximum adsorption capacity obtained from the isotherms shown in Fig. 4. Also, the error associated for all the calculated Q0 values is less than 0.4, as determined by using the confidence interval of 95% and the tStudent distribution. The Q0 values calculated by us for carbomix agree very well with the reported values by Atta-Politou et al., 2001. These authors used potentiometry and a fluoxetine ion selective electrode to study the adsorption of fluoxetine by carbomix at pH 1.2 and reported the value 405 mg/g for Q0. Only a few papers were published about the study of fluoxetine adsorption. Nevertheless, we can favourable compare our calculated Q0 for the activated carbon samples with the values reported by Kamp et al. [16] and Cooney and Thomason [17]. Table 2 also shows some other relevant data in the adsorption process. Carbon materials have amphoteric characteristics due to the coexistence of basic and acidic functional groups on the carbon surface (in our case the concentration of both groups can be seen in Table 1). The nature and concentration of the charge on the carbon surface are determined by the solution pH and the material pzc value. The relation between these two pH values controls the ionisation of the surface functional groups which, in turn, establish the electric mean charge of the material surface. In general terms we can observe a positively or negatively charged surface when the solution pH is inferior or superior to the material pzc, respectively [18]. The charge intensity is proportional to the difference between the solution pH and the pzc value. On the other hand, the fluoxetine HCl, showed in Fig. 1, has pKa value equal to 9.5 which means that at the working pH values, 1.2 and 7.5, about 100 and 90% of the fluoxetine molecules are respectively in the protonated form. Recently two review papers were published about the adsorption of organic molecules onto carbon materials [18,19] where the authors concluded that the adsorption mechanisms of a number of substances are not fully known yet. Nevertheless, the principal adsorption mechanisms referred to in

Fig. 4 – Fluoxetine adsorption isotherms at 37 °C.

554

F U E L P RO CE SS I NG T EC H NOL O G Y 8 9 (2 0 0 8) 5 49 –5 5 5

these papers mainly involve the influence of electrostatic interactions, specific chemical interactions between the adsorbate molecule and functional groups on the material surface and dispersive interactions between the delocalised electrons at the carbon basal sites and the aromatic structure of the adsorbate. Additionally, some considerations about the possibility of molecular sieving effects were discussed. The data in Table 2 clearly shows that, in most cases, the adsorption capacity at pH 7.5 is higher than at pH 1.2. However, we can observe different behaviour for the activated carbon and the activated carbon fibres. In the first case the enhanced adsorption at higher pH can be due to more favourable electrostatic interaction, as indicated by the carbon materials and fluoxetine electric charge exposed in Table 2. On the contrary, sample F953ox shows an opposite behaviour. Despite the more favourable electrostatic interactions at pH 7.5, with the existence of some electrostatic attraction between the positively charged fluoxetine molecules and the negatively charged ACF surface, the adsorbed quantity was smaller than for pH 1.2. This fact indicates that in this case the electrostatic interactions did not have a relevant impact on the adsorption mechanism. The same general conclusion can be drawn if we compare the results obtained by F953 and F953ox at pH 1.2. Samples F953 and F953ox adsorb almost the same amount of fluoxetine, Q0 for F953 is 103 mg/g and for F953ox is 97 mg/g, and have very similar porous characteristics, as shown in Table 1, but different chemical properties, as pointed out by the different pzc values showed in Table 1. This difference leads to less important electrostatic repulsion in the F953ox sample due to a small difference between solution pH and the pzc value, which produces a less intense positive charge on the ACF surface. It seems that in the ACF case the most relevant factor for the adsorption of fluoxetine is the molecular sieving effect due to the narrower pore size distribution and to the existence of pores with inferior mean pore width, when compared with the activated carbon samples used. Also, it is possible that dispersive interactions can also have some relevancy. The adsorption capacity decrease found for F953ox at pH 7.5 can be due to a higher localisation of the π electrons at the basal ACF planes. This localisation is caused by the surface groups' ionisation and has as consequence less intense interactions with the aromatic molecule of fluoxetine. Nevertheless, we think that in order to increase the capacity for the adsorption of biomolecules, such as fluoxetine, we need to develop some mesoporosity in the materials by changing the production method or by pos-production treatments. If we compare the results obtained for carbomix carbon before and after the treatments we can observe that the oxidation leads to an increase of the capacity for the adsorption of fluoxetine of about 30%. In contrast, the thermal treatment results in a decrease of the adsorption capacity of approximately 15%. As already noted the treatments only cause slight changes on porosity but significant alterations of the surface chemistry, oxidation leads to the formation of an acidic surface, where the electrostatic interactions for the fluoxetine adsorption are more favourable, and the thermal treatment to a increment of the pzc value from 8.01 to 9.52, and thus to a surface where the predominance of basic groups involves more intense electrostatic repulsions between the carbon sur-

face and the fluoxetine molecules. Therefore, we can remark that in the carbomix carbon the most relevant factor in the adsorption mechanism probably is the electrostatic interactions. The same general conclusion can be obtained for maxsorb carbon. Our results shows that for the activated carbon samples with similar pzc values, carbomix (pzc = 8.01) and maxsorb (pzc = 8.30), the textural properties are determinant to the fluoxetine adsorption ability and we can observe a direct relation between the pore volume, or BET apparent surface area, and the maximum adsorption capacity of the carbons. As an example, the values of ABET and Vs for carbomix are approximately half those for maxsorb and, for each pH, the same relations occur for the calculated Q0 values, as can be seen in Table 2. Kamp et al. [16], reported that the functional groups carbonyl and carboxyl can act as adsorption sites for fluoxetine due to enhanced chemical interactions. Our results can also be explained by this chemical interaction because it is well known that the oxidation by nitric acid in liquid phase will increase the amount of carbonyl and carboxyl group [20,21]. The use of ordered mesoporous materials for bioadsorption and biocatalysis has been recently reviewed by Hartmann [22]. These materials have a significant impact on the adsorption and immobilisation of proteins indicating a favourable use as adsorbent for large molecules, such as fluoxetine. In this work we used Ti– and Al–MCM-41 materials synthesised in our laboratory. The MCM-41 material is the most widely studied member of the M41S family disclosed, 15 years ago, by the scientists of the Mobil Corporation Strategic Research Center [23,24]. The structure of MCM-41 material consists in a hexagonal array of unidirectional tubular pores. The important features are the extremely high surface area and porosity, narrow pore size distributions and pore size adjustable from ∼2 to 10 nm. In fact, these unique properties combined in one material, make it highly attractive for a wide range of potential applications including adsorption as mentioned above. Despite the high potential characteristics, exploratory trials reveal that the use of the Ti–MCM-41 materials have, at this point of the work, a smaller adsorption capacity than the activated carbons tested but similar adsorptive properties to the ACF sample. For example, adsorbed amounts of 15 and 41 mg/g respectively at equilibrium concentrations of 0.5 and 2 mmol/L were observed for TiBe16-50 at pH 1.2. Therefore, we can conclude that there is a need to improve the adsorption by these materials and the preparation of functionalised materials may be a possible strategy to reach our goal. Nevertheless, it should be mentioned that a considerable increase to ∼ 100 mg/g for the concentration of 0.5 mmol/L was observed for TiEp18-5 by increase of pH to 7.5. As previously noticed, higher titanium content samples present some acidity mainly due to the extra framework presence of partial polymerized titanium species [8]. Furthermore the most acid MCM-41 sample tested, Al–MCM-41, allows us to obtain an adsorbed amount of ∼ 210 mg/g for the concentration of 0.1 mmol/L at pH = 7.5, which is reasonably high and comparable to some of the carbon materials. So, although further experiments must be carried out in order to clarify the adsorption mechanism, it appears that surface acidity may play a role in enhancing the adsorption of fluoxetine by these mesoporous type materials.

F U E L P R O CE SS I NG T EC H NOL O G Y 8 9 (2 0 0 8 ) 5 4 9–5 5 5

4.

Conclusion

If a person was to ingest an entire box of 50 Prozac® capsules of 20 mg and this was to mix with 1 L of stomach fluid, a drug concentration of 1 g/L would result, which is in the range of an expected overdose. The normal dose of activated carbon slurry to treat overdoses involves the use of about 60 g of activated carbon. Taking into account the maximum adsorption capacity of the materials tested by us and the above data we can consider that, despite the differences between them, all the materials have potential to be used to treat fluoxetine intoxications since a dose of 60 g will absorb more than 1 g of fluoxetine. We can also conclude that the activated carbon fibres have a distinct behaviour to that of the activated carbons tested, namely in the principal adsorption mechanism and the influence of the electrostatic interactions. It seams that in the first case the most relevant factor is the carbon molecular sieve effect and interactions of dispersive nature between the ACF basal plane and the fluoxetine molecule. On the other hand, in the activated carbons it seams that the adsorption is much more dependent on the electrostatic interactions between the carbon charged surface and the protonated fluoxetine molecule. In general terms we can also conclude that the adsorption of fluoxetine is enhanced in carbon materials with acidic characteristics.

[8]

[9]

[10]

[11]

[12] [13]

[14]

[15] [16]

Acknowledgements The authors are grateful to the Fundação para a Ciência e a Tecnologia (Portugal) and the European Regional Development Fund (FEDER) for financial support. We are grateful to Eli Lilly Portugal for providing pure drug compound for developing these assays and to Norit N.V. and Kansai Coke & Chemicals Co for the provision of activated carbon samples.

[17] [18]

[19] [20]

REFERENCES [1] M.S. Milane, M.A. Suchard, M.L. Wong, J. Licinio, Modeling of the temporal patterns of fluoxetine prescriptions and suicide rates in the United States, PLoS Med. 3 (6) (2006) e190, doi:10.1371/journal.pmed.0030190. [2] WHO, The world health report 2001 — mental health: new understanding, new hope, www.who.int/entity/whr/2001/en/ whr01_en.pdf2001. [3] R. Mandrioli, V. Pucci, D. Visini, G. Varani, M.A. Raggi, Rapid methods for determination of fluoxetine in pharmaceutical formulations, J. Pharm. Biom. Anal. 29 (2002) 1127–1134. [4] J. Atta-Politou, I. Skopelitis, I. Apatsidis, M. Koupparis, In vitro study on fluoxetine adsorption onto charcoal using potentiometry, Euro. J. Pharm. Sci. 12 (2001) 311–319. [5] D. Seger, Position paper: single-dose activated charcoal, Clinical Toxicology 43 (2005) 61–87. [6] P.J.M. Carrott, J.M.V. Nabais, M.M.L. Ribeiro Carrott, J.A. Pajares, Preparation of activated carbon fibres from acrylic textile fibres, Carbon 39 (2001) 1543–1555. [7] C. Galacho, M.M.L. Ribeiro Carrott, P.J.M. Carrott, Evaluation of the thermal and mechanical stability of Si–MCM-41 and

[21]

[22] [23]

[24]

[25]

555

Ti–MCM-41 synthesised at room temperature, Micropor. Mesopor. Mater. (in press) doi:10.1016/j. micromeso.2007.04.10. C. Galacho, M.M.L. Ribeiro Carrott, P.J.M. Carrott, Structural and catalytic properties of Ti–MCM-41 synthesised at room temperature up to high Ti content, Micropor. Mesopor. Mater. 1-3 (2007) 312–321. M.M.L. Ribeiro Carrott, F.L. Conceição, J.M. Lopes, P.J.M. Carrott, C. Bernardes, J. Rocha, F. Ramôa Ribeiro, Comparative study of Al–MCM materials prepared at room temperature with different aluminium sources and by some hydrothermal methods, Micropor. Mesopor. Mater. 92 (2006) 270–285. J. Rouquerol, D. Avnir, C.W. Fairbridge, D.H. Everet, J.H. Haynes, N. Pernicone, J. Ramsay, K.S. Sing, K.K. Unger, Recommendations for the characterization of porous solids, Pure and Applied Chemistry 66 (8) (1994) 1739–1758. P.J.M. Carrott, R.A. Roberts, K.S.W. Sing, Standard nitrogen adsorption data for nonporous carbons, Carbon 25 (6) (1987) 769–770. F. Rouquerol, J. Rouquerol, K.W. Sing, Adsorption by Powders & Porous Solids, Academic Press, London, 1999. P.A. Russo, M.M.L. Ribeiro Carrott, A. Padre-Eterno, P.J.M. Carrott, P.I. Ravikovitch, A.V. Neimark, Interaction of water vapour at 298 K with Al–MCM-41 materials synthesised at room temperature, Micropor. Mesopor. Mater. 103 (2007) 82–93. P.J.M. Carrott, J.M.V. Nabais, M.M.L. Ribeiro Carrott, J.A. Menendez, Thermal treatments of activated carbon fibres using a microwave furnace, Micropor. Mesopor. Mater. 47 (2001) 243–252. C.H. Giles, T.H. Ewan, S.N. Nakhna, S. Smith, Studies in Adsorption. Part XI, J. Am. Chem. Soc. (1960) 3973–3993. K.A. Kamp, D. Qiang, A. Aburub, D.E. Wurster, Modified Langmuir-like model for modelling the adsorption from aqueous solutions by activated carbons, Langmuir 21 (2005) 217–224. D.O. Cooney, R. Thomason, Adsorption of fluoxetine HCl by activated charcoal, J. Pharm. Sci. 86 (5) (1997) 642–644. L.R. Radovic, C. Moreno-Castilla, J. Rivera-Ultrilla, Carbon materials as adsorbents in aqueous solutions, in: L.R. Radovic (Ed.), Chemistry and Physics of Carbon, vol. 27, Marcel Dekker Inc., New York, 2000, pp. 228–405. C. Moreno-Castilla, Adsorption of organic molecules from aqueous solution on carbon materials, Carbon 42 (2004) 83–94. C.Y. Yin, M.K. Aroua, W.M. Daud, Review of modifications of activated carbon for enhancing contaminant uptakes from aqueous solutions, Separation and Purification Technology 52 (2007) 407–415. J. Trawczynski, S. Suppan, C. Sayag, G. Dlega-Mariadassou, Surface acidity of the activated CBC, Fuel Processing and Technology 77-78 (2002) 317–324. M. Hartmann, Ordered mesoporous materials for bioadsorption and biocatalysis, Chem. Mater. 17 (2005) 4577–4593. C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J.S. Beck, Ordered mesoporous molecular sieves synthesized by liquid–crystal template mechanism, Nature 359 (1992) 710–712. J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmidt, C.T.-W. Chu, D.H. Olson, E.W. Sheppard, S.B. McCullen, J.B. Higgins, J.C. Schlenker, A new family of mesoporous molecular sieves prepared with liquid crystal template, J. Am. Chem. Soc. 14 (1992) 10834–10842. P.J.M. Carrott, P.A.M. Mourão, M.M.L. Ribeiro Carrott, E.M. Gonçalves, Separating surface and solvent effects and the notion of critical adsorption energy in the adsorption of phenolic compounds by activated carbons, Langmuir 21 (2005) 11863–11869.