e-Polymers 2006, no. 081

http://www.e-polymers.orgISSN 1618-7229

Thermal Oligomerisation of Cardanol Francisco Helder A. Rodrigues,1 José Roberto R. Souza,2 Francisco Célio F. França,2 Nágila M. P. S. Ricardo,2 Judith P. A. Feitosa2*

1Coordenação de Química, Universidade Estadual Vale do Acaraú, Sobral, Ceará, Brasil; 2Departamento de Química Orgânica e Inorgânica, Campus do Pici, Universidade Federal do Ceará, Fortaleza, Ceará, Brasil; Fax: 00-55-85-33669978; judith@dqoi.ufc.br (Received: 06 February, 2006; published: December 26, 2006)

Abstract: Cardanol was extracted from technical Cashew Nut Shell Liquid (CNSL), a naturally occurring meta-substituted long chain phenol and oligomerized by heating at 140 °C. Products were characterized by rheology, infrared and 1H NMR spectroscopy and thermogravimetric analysis (TGA). Increase in viscosity and the flow activation energy was found with increasing time of heating. The relative absorbance of double bond of the hydrocarbon chain decreased with time of heating and indicated that the oligomerization is taking place through the unsaturation of the side chain. Decrease in internal double bond as well as in vinyl bond, observed by 1H NMR, pointed out to the participation of these two kinds of unsaturation as well as monoene, diene and triene. The oligomerization is a slow process. With 40 h of heating, the average molecular weight increases only by 46%. Great differences were found in TGA curves of cardanol with different times of oligomerization. Thermal stability increases with time of heating. The degree of oligomerization could be determined from relative mass loss of the first event of TG, or from flow activation energy.

Introduction Cashew nut shell liquid (CNSL) is a unique natural source for unsaturated long-chain phenols. It is a cheap and renewable material, obtained as a by-product of the cashew industry. CNSL by itself is useful as insecticidal, fungicidal, anti-termite and for medicinal applications, but its main applications are in the polymer chemistry, where it can replace phenol in many instances with equivalent or better results [1]. The major phenolic constituents of CNSL are cardol, anacardic acid and cardanol. The technical CNSL, which is obtained by roasting the cashew shell at 180-200°C, does not contain anacardic acid, because it is thermolabile and converted to cardanol [2]. Cardanol, a main component of technical CNSL, is a monophenol with a C15 unsaturated side chain at the meta-position, mainly with 1-3 double bond (Figure 1) [3]. It has a potential use in resins, friction lining materials, and surface coatings [4]. Simple derivatives of cardanol were reported to have industrially important properties [2]. Chlorinated products were found to have pesticidal action. Sulfonated derivatives are used as surface-active agents. They are also applied in the form of dyestuffs, plasticizing and ion-exchange resin. In recent international patent base, cardanol and derivatives are also mentioned to be used in many applications [5-9]. They are used

1

as modifier in coatings, adhesives, sealants, rubbers, plastics, elastomers, composites or inks [5], in cosmetics for skin [6], in mosquito abatement [7], to coat granules used for fertilisers, insecticides, plant protection agents, fungicides and drying agent [8] and friction material for brake pad [9].

OH

R

8'

8'

8'

11'

11' 14'

n = 0

n = 1

n = 2

n = 3

5.4%

48.5%

16.8%

29.3%

R = C15H31- 2n

Fig. 1. Structure of cardanol with side chains [3]. Another field of application for cardanol and derivatives is as antioxidants and stabilisers in fuels [10] and in polyisoprene rubber [11]. They protect rubber from thermal degradation at 140 °C, which is the temperature used in vulcanization process [11]. Rodrigues et al. [11] have proposed that the oligomerization or polymerization of CNSL, cardanol and derivatives could have influence over their antioxidant effect. Low molecular weight antioxidants are easily lost from polymer through migration, evaporation, and extraction. The presence of polymerized material could decrease the migration of the antioxidant to the surface, reduce its volatilization and retain the antioxidant activity during heating. The possibility of the presence of polymerised material in technical CNSL is well known [2]. Cardanol can be polymerized either by condensation with aldehydes or by chain reaction polymerization using chain unsaturation [12]. Polycardanol has been obtained through enzyme-catalysed polymerization by the use of soybean [13] and fungal peroxidases [14], and also horse radish peroxidase in the presence of redox mediators [15]. Even though cardanol has two groups subject to polymerization (phenolic moiety and unsaturated hydrocarbon group) only phenolic moiety was polymerised during these enzymatic catalysed reactions [13-15]. The only study of polymerization/oligomerization through the side chain was reported by Manjula et al. [16] using acid catalysts. The aim of this work is to study the kinetics of thermal oligomerization of cardanol without any kind of catalyst in order to make it possible the evaluation of the

2

contribution of the oligomerized material over properties of cardanol, in systems where it has to be heated, for example, in thermal oxidation of polyisoprene. Results and discussion Rheology Cardanol samples were heated at 140 °C for different times (5, 10, 15, 20, 30 and 40 h). The viscosity increases with time of heating. Samples heated during 30 and 40 h have very high viscosity, impossible to be measured with the spindle used for other samples. The effect of time of heating and measurement temperature over absolute viscosity of cardanol heated up to 20 h is shown in Fig. 2. The viscosity increase is 13 and 7 times compared with the unheated sample for measurement made at 10 and 50 °C, respectively.

0 5 1 0 1 5 2 00

2

4

6

8

1 0 ºC 2 0 ºC 3 0 ºC 4 0 ºC 5 0 ºC

visc

osity

[Pa.

s]

t im e o f h e a tin g [h ]

Fig. 2. The effect of time of heating at 140°C and measurement temperature over absolute viscosity of cardanol. Doolitle [17] presented a simple and successful correlation between the viscosity and the average molecular weight of homologous series: η = a MWb (1) where a and b are constants for a given series and MW is the molecular weight of a homologue. Rønningsen [18] showed that at constant temperature, this simple relationship applies quite well even to highly complex mixtures such as crude oils. The augment of viscosity of cardanol with time of heating could be attributed to molecular weight increase, indicative of oligomerization. Manjula et al. [16] proposed that the oligomerization of cardanol by heating at 180 °C during 30 min occurs with the formation of 2.68% dimer and 3.19% trimer. Tetramer was formed in reaction using acid catalysts. Flow activation energy (Ea) is the energy necessary to overcome a potential barrier hindering the displacement of the particles from an equilibrium point to another

3

during the shear of the sample. It could be calculated by the Arrhenius-Frenkel-Eyring equation [19]:

η = B exp(Ea/RT) (2)

where B is a constant and T the temperature in Kelvin. Linear dependence of lnη vs T-1 was observed for all heated cardanol. The calculated Ea depends on time of heating (Fig. 3).

0 5 10 15 20 2544

48

52

56

60

E a [kJ

.mol

-1]

time of heating [h]

Fig. 3. Effect of time of heating on the flow activation energy (Ea) of cardanol. A linear correlation between flow activation energy and molecular weight was found for a series of n-alkanes and petroleum oils [18]. Siline and Leonov [20] correlated the flow activation energy of monomer fluid, E1, with the flow activation energy on chain length N, EN, for the polymer by the eq. 3: EN = E1. n (3) where n is the average length of the flow segment at a given degree of polymerization. Taking into account that the flow segment for oligomers could be their molecular weight and combining equation 3 with considerations of Rønningsen [18] about the dependence of Ea and molecular weight, equation 4 could be proposed: Ea(t) / Ea (0) = MW(t) / MW(0) (4) where Ea(t) and Ea(0) are flow activation energies at time of heating t and of unheated cardanol, respectively. In the same way, MW(t) and MW(0) are average molecular at time of heating t, and of unheated cardanol, respectively. Based on the composition of cardanol [3], MW(0) is 300.6. Using the linear correlation from Fig. 3 (Ea = 45.31 + 0.52 t, r2 = 0.995) and eq. 4, the MW of heated cardanol for 5, 10, 15 and 20 h are 318.6, 333.7, 351.7 and 360.7, respectively. By extrapolation, average MW for 30 and 40 h could be 402.8 and 438.6, respectively. The intrinsic viscosity of cardanol heated for 30 h was determined in toluene at 40 °C and compared with values reported by Manjula et al. [16]. Based on the correlation of [η] with the composition of oligomer obtained from acid-catalysis of cardanol, a

4

tentative composition was proposed. Thermal oligomerization of cardanol at 140 °C for 30 h produces a material with [η] of 0.011 dl/g and composed by approximately 75% of monomer, 8% of dimer, 5% of trimer and 7% of tetramer. This composition gives average MW of 402.8, in agreement with that estimated from flow activation energy. Infrared spectroscopy FT-IR spectra for unheated (0) and heated cardanol for 20 and 40 h are depicted in Fig. 4. The spectra of unheated sample present the typical absorption of cardanol [16]. After heating loss of unsaturation in side chain could be observed. A decrease in absorbance at 3008, 1635 and 908 cm-1, attributed to C-H stretching from -CH=CH- or -CH=CH2 groups, C=C stretching and CH out of plane deformation of terminal vinyl, respectively, confirms this assumption [21]. There is no reduction on the relative absorbance at 3370 cm-1, attributed to stretching vibration of phenolic OH. The oligomerization is taking place through the double bonds of the side chain, as occurred in acid-catalysed reaction [16].

4 0 0 0 3 5 0 0 3 0 0 0 2 5 0 0 2 0 0 0 1 5 0 0 1 0 0 0 5 0 0

Wavenumber [cm-1]

Abs

orba

nce

0

40 h

20 h

Fig. 4. FT-IR spectra of unheated and heated cardanol for 20 and 40 h. The kinetics of the thermal oligomerization could be observed in Fig. 5. An internal standard was used to calculate the relative absorbance. For the band at 3008 cm-1 (ν C-H of CH=CH) the internal standard was the band at 2856 cm-1, due to νC-H vibration of saturated segments of side chain (CH2 and CH3), as done by Ikeda et al. [22]. For the band at 1635 cm-1 (νC=C of side chain) the band at 1592 cm-1, ascribed to νC=C of aromatic ring, was used as done by Gauthier at al. [23]. The dependence of the relative absorbance at 3008 and 1635 cm-1 with time of heating is exponential and agrees with the first order kinetics determined by Manjula et al. [16] for the oligomerization of cardanol using acid catalysts.

5

0 10 20 30 400.0

0.2

0.4

0.6

0.8

time of heating [h]

Abs

rel a

t 300

8 cm

-1

0.0

0.1

0.2

0.3

0.4

Absrel at 1635 cm

-1

Fig 5. Evolution of relative absorbance of =C-H and C=C stretching of cardanol with time of heating. 1H NMR spectroscopy NMR study was done in order to determine the participation of internal double bond and vinyl bond from the side chain in the oligomerization process. NMR spectra of unheated cardanol and the sample heated for 20 and 40 h were depicted in Fig. 6 and the main attributions are summarized in Table 1. It could be observed in Fig. 6 that some peaks increase in intensity and some decrease. They can be separated in signals from vinyl or terminal protons, proton bonded to internal -C=C-, and aromatic protons. Considering vinyl or terminal protons, there is a decrease of intensity of peaks at 5.8 (-CH=) and 5.1 ppm (=CH2) and an augment of peak at 0.98 ppm (CH3), attributed to cleavage of C20' double bond in the triene (Tab. 1) to form the oligomer. The oligomerization is also occurring from internal double bond (C14' and C17' or triene, and/or C14' in diene), as verified by the intensity diminishing of peaks at 5.4-5.5 (-CH=CH-) and the augment of peak at 1.4 ppm, assigned to (CH2)n. Relative area of signals taking aromatic ring proton as standard (6.7-7.2 ppm) was plotted against time of heating. Fig. 7 is related to loss of internal unsaturation and Fig. 8 to vinyl bond. There is a reduction in -CH=CH- and a concomitant increase in (CH2)n (Fig. 7). The reduction in internal double bond due to oligomerization for 30 and 40 h of heat is 16 and 19%, respectively. In the same way the decrease in -CH= and =CH2 from vinyl group was associated with the increase in terminal CH3 (Fig. 8). For cardanol heated for 30 and 40 h, 37 and 46% of vinyl bond were lost.

6

20 h

40 h

01 234 56 7 8 ppm

Fig. 6. 1H NMR spectra of unheated cardanol and heated for 20 and 40 h. Tab. 1. Assignments of 1H NMR signals of cardanol.

14' 15'

1 6'

17' 18'

19'

20' 21'

14' 15'

1 6'

17' 18'

19''

20

21

14' 15'

16''

17

18

19

20

21

14

15

16

17

18

19

20

21R =

R

OH

7

8

9

10

11

12

13''

1 2 3

4 5 6

entry group δ [ppm]

20’ -CH= 5.8 14’-15’ 17’-18’ -CH=CH- 5.4-5.5

21’ =CH2 5.1 5 (CH)Ar 7.2 4 (CH)Ar 6.8

2 and 6 (CH)Ar 6.7 16’-16’’ 19’-19’’ CH2CH=CH- 2.9

7 ArCH2 2.6 13’’ CH2CH=CH- 2.0 8 ArCH2CH2 1.6

9-12 (CH2)n 1.4 21 CH3 0.98

7

0 10 20 30 400.6

0.7

0.8

0.9

1.0

time of heating [h]

1 H in

tegr

al o

f -C

H=C

H-

3

4

5

6

1H integral of (C

H2 )n

Fig. 7. Evolution of relative area of -CH=CH- and (CH2)n in 1H NMR spectra of cardanol with time of heating (internal double bond oligomerization).

0 10 20 30 400.1

0.2

0.3

0.4

time of heating [h]

1 H in

tegr

al o

f -C

H a

nd =

CH

2

0.2

0.4

0.6

0.8

1H integral of C

H3

Fig. 8. Evolution of relative area of -CH and =CH2 in 1H NMR spectra of cardanol with time of heating (vinyl double bond oligomerization). Taking into consideration the composition suggested by rheology, which indicates the dimer as the main oligomer that was formed, two possible dimerization reactions are presented in Fig. 9: (I) obtained from internal double bond loss, taking the monoene as example, and (II) from vinyl loss in triene. There are other possibilities for dimer structure and many other for trimer and tetramer. In reaction I the decrease in the number of double bond and increase in (CH2)n can be visualised. In reaction II the increase in number of CH3 group due to vinyl bond loss is evident.

8

OH

OH

OH

Δ

I

OH

OH

OH

Δ

II

Fig. 9. Example of dimerisation reaction of cardanol from: (I) internal double bond loss, taking the monoene as example, and (II) vinyl loss in triene. Thermogravimetric analysis TGA curves of unheated and heated cardanol was obtained under air atmosphere in order to investigate the thermal stability. In all cases, a three-step decomposition was observed (Fig. 10 and 11). In the first step (at 160-360 °C) oligomerization and oxidative decomposition could be some of the occurring processes. The second and third steps (at 390-550 °C) involve a process of decomposition which gives rise to a complete volatilization. The mechanism of thermo-oxidative decomposition is probably complex, considering the initial mixture of structures in cardanol (Fig. 1) and the different possibilities of oligomer structures obtained by the previous heating. Relevant parameters of thermal stability from TGA curves were given in Tab. 2. There is a clear shift of the initial decomposition temperature (Ti) towards higher temperature with time of previous heating. On the opposite direction is the shift of the end temperature of decomposition (Tf). Similar behaviour was observed by Lai et al. [24] with p-dioxanone prepolymer (PPDO) in comparison with chain-extended PPDO. The maximum decomposition temperature (Tmax) of event I did not change substantially with oligomerization and stayed in the range of 270-280 °C. The variation on Tmax of events II and III between heated material and cardanol reaches 30 and 27 °C, respectively, but with no obvious trend. The residue at 800 °C is around 2% for all materials. The most evident change with oligomerization is in the mass loss for the first event and temperature of 50% of mass loss (T50) (Tab. 2). Cardanol lost 65.2 % of mass in the first event, while the oligomerized by 40 h presented only 30.5% of mass loss for the same event. Shift on mass loss in the first stage of thermal decomposition was obtained when benzoxazine trimer was compared with tetramer, 80 and 70 %, respectively [25]. The T50 of cardanol oligomer (40 h of heating) shifts 149 °C to higher temperature, in comparison with monomer. Shift in the same order of magnitude (180 °C) was reported to oligomer of 2-[(4-fluorophenyl) imino methylene] phenol in relation with its monomer [26].

9

0 200 400 600 800

0

20

40

60

80

100

resi

dual

mas

s [%

]

temperature [ºC]

0

20 h

40 h

Fig. 10. Thermogravimetric curves for the unheated cardanol and heated for 10 and 40 h. TG in synthetic air atmosphere and rate of heating of 10 ºC/min.

0 200 400 600 800

DTG

[mg/

min

]

temperature [ºC]

40 h

20 h0

Fig. 11. DTG for the unheated cardanol and heated for 20 and 40 h. TG in synthetic air atmosphere and rate of heating of 10 ºC/min.

10

Tab. 2. Parameters of the thermal degradation process of the unheated and heated cardanol at different times (TGA condition: synthetic air and heating rate = 10 ºC/min).

times of cardanol heating in h parameter unheated

cardanol 5 10 20 30 40

Ti in °C 165 167 169 174 177 181

Tf in °C 652 621 618 534 531 506

I 270 283 282 282 280 279

II 427 456 427 455 457 441 Tmax in °C

III 473 479 458 461 510 484

mass loss on event I in % 65.2 59.5 52.0 46.2 40.2 30.5

T50 in °C 292 302 318 386 425 442

IPDTa in °C 293 301 316 326 339 359 aIntegral procedure decomposition temperature

0.4 0.5 0.6 0.7 0.8 0.90.9

1.1

1.3

1.5

1.7

DO

relative Δm

Fig. 12. Relation between the degree of oligomerization of cardanol and the relative mass loss of the first event on thermogravimetric curve. Taking into account the changes in Ti, T50, and IPDT, the oligomerization improves the thermal stability which increases with the time of heating. The degree of oligomerization (DO) could be defined in the same way as the degree of polymerization (DP). DO = MW(oligomer)/ MW(monomer) (5) Based on eq. 4, DO = Ea(oligomer)/ Ea(monomer) (6)

11

Mass loss of the first event of thermal decomposition (Δm) varies with time of heating (Tab. 2). As the degree of oligomerization also vary with time of heating, a relationship between DO and Δm was searched. Considering that DO is a relative parameter, a relative Δm was used, defined as Δmoligomer/Δmmonomer. Fig. 12 shows that the degree of oligomerization increases with the decrease in relative Δm. In the intervals of time of heating from 10 to 40 h (relative Δm from 0.5 to 0.8) a linear correlation was found (r2 = 0.992). TGA could be an easy way for the determination of the degree of oligomerization of cardanol. The following equation is proposed:

DO = 1.98 - 1.08(Δmoligomer/Δmmonomer) (7) Conclusions Cardanol could be thermal oligomerized at 140 °C. The reaction took place through the unsaturation of side chain, both from internal double bond (monoene, diene and triene) and vinyl bond (triene). The oligomerization was a slow process. With 40 h of heating, the average molecular weight increases only 46%. Increase in viscosity and flow of energy of activation was found with time of heating. Great differences were found in thermogravimetric curves of cardanol with different times of oligomerization, which points to a increase on thermal stability with time of heating. The degree of oligomerization could be determined from relative mass loss of the first event of TG, or from flow activation energy. Experimental Materials Cardanol was obtained from technical cashew nut shell liquid (CNSL) donated by the cashew nut processing plant CIONE (Fortaleza-CE). The isolation procedure of Kumar et al. [2] was followed and is briefly described here. CNSL was dissolved in methanol and ammonium hydroxide (Reagen) and stirred. This solution was then extracted with hexane and the organic layer washed with HCl followed by distilled water. Activated charcoal was added and the organic layer filtrated and concentrated to give pure cardanol. All reagents and solvents, except NH4OH, were supplied from Synth. They were used without further purification. Heating The materials were heated up in Petri plates of 5 cm of diameter by the use of a warming chamber Model-19 (Thelco), at constant temperature of 140 ± 1 °C and in atmospheric air for varied times. Initial mass of 2.0 g was always used. Rheology A Brookfield model DV-III equipment with spindle CP-52 was used for the viscosity measurements. The measurements were accomplished in several temperatures, controlled by a thermostatic bath with accuracy of 0.1 °C. Intrinsic viscosity was determined in toluene at 40 °C by the use of an Ubbelohde viscometer with capillary diameter of 0.5 mm. Flow time of solvent was 80 s. All flow time was measured at least three times to calculate the average values.

12

Infrared analysis Infrared spectra were carried out using an equipment Shimadzu FTIR–8300, in absorbance mode, with the material in KBr pellets. 1H NMR NMR spectra were recorded at ambient temperature using a Bruker Model Avance DRX-500 spectrometer, operating at 500 MHz. CDCl3 99.8% from Cambridge Isotope Laboratories was used as solvent. Thermogravimetric analysis TGA curves were carried out in synthetic air using a Shimadzu TGA-50 instrument at a heating rate 10 ºC/min over a temperature range of 25-800 °C. The initial mass of sample was maintained in 10 mg and flow rate of air in 50 ml/min. Acknowledgements The authors acknowledge CIONE for the donation of cashew nut shell liquid (CNSL) and CENAURMN for the NMR spectra. The authors would like to thank CAPES and CNPq for the financial support. References [1] Lubi, M. C.; Thachill, E. B.; Desig. Mon. and Polym. 2000, 3, 123. [2] Kumar, P. P.; Paramashivappa, R.; Vithayathil, P. J.; Rao, P. V. S.; Rao, A. S.; J. Agric. Food Chem. 2002, 50, 4705. [3] Tyman, J. H. P.; Chem. Soc. Ver. 1979, 499, 8. [4] Menon, A. R. R.; Pillai, C. K. S.; Sudha, J. D.; Mathew, A. D.; J. Sci. Ind. Res. 1985, 44, 324. [5] Dia, Z.; Chen, M. J.; Daí, Z.; U.S. Patent 2001, 6229054-B1; Canadian Patent 2002, 1332712-A. [6] Kanebo LTD; Japanese Patent 2000, 0204030-A. [7] Hammond, D.G.; Kubo, I.; U.S. Patent 2000, 6077521-A. [8] Winter R.; Priebe, C.; Kuhlmann, P.; Germany Patent 2003, 048075-A; U.S. Patent 2005, 005661-A1; Spanish Patent 2005, 2242079-T3. [9] Baba, T.; Aoyagi, Y.; Sasaki, Y.; Japanese Patent 2005, 233214-A. [10] Dantas, T. N.; Dantas, M. S. G.; Dantas Neto, A. A.; D'Ornellas, C. V.; Queiroz, L. R.; Fuel 2003, 82, 1465. [11] Rodrigues, F. H. A.; Feitosa, J. P. A.; Ricardo, N. M. P. S.; de França, F. C. F.; Carioca, J. O. B.; J. Braz. Chem. Soc. 2006 , 17, 265. [12] Murthy, B. G. K.; Menon, M. C.; Agarwal, J, S.; Zaheer, S. H.; Paint Manuf. 1961, 31, 47. [13] Ikeda, R., Tanaka, H.; Uyama, H.; Kobayashi, S.; Polym. J. 2000, 32, 589. [14] Kim, Y. H.; Won, K.; Kwon, J. M.; Jeong, H. S.; Park, S. Y.; An, E. S.; Song, B. K.; J. of Mol. Catal. B: Enzymatic 2005, 34, 33. [15] Won, K.; Kim, Y. H.; An, E. S.; Lee, Y. S.; Song, B. K.; Biomacromolecules 2004, 5, 1. [16] Manjula, S.; Kumar, V. G.; Pillai C. K.; J. Appl. Polym. Sci. 1992, 45, 309. [17] Doolitle, A. K.; J. Appl. Phys. 1952, 23, 418. [18] Rønningsen, H. P.; Energy Fuels 1993, 7, 565.

13

[19] Vinogradov, G. V.; Malkin, A. Y.; "Rheology of Polymers, Viscoelasticity and Flow of Polymers", Mir, Moscow 1980, pp. 105-121. [20] Siline, M.; Leonov, A. I.; Polymer 2002, 43, 5521. [21] Bellamy, L. J.; "The Infrared Spectra of Complex Molecules", 3rd ed.,Chapman and Hall, London 1975, pp.13-50. [22] Ikeda, R.; Tanaka, H.; Uyama, H.; Kobayashi, S.; Polymer 2002, 43, 3475. [23] Gauthier, M. A.; Stangel, I.; Ellis, T. H.; Zhu, X. X.; Biomaterials 2005, 26, 6440. [24] Lai, Q.; Wang, Y.-Z.; Yang, K.-K.; Wang, X.-L.; Zeng, Q.; React. Funct. Polym. 2005, 65, 309. [25] Hemvichian, K.; Kim, H. D.; Ishida, H.; Polym. Degrad. Stab. 2005, 87, 213. [26] Kaya, I.; Gül, M.; Eur. Polm. J. 2004, 40, 2025.

14