J. Phys. Chem. B 2006, 110, 8047-8051
8047
Physical Insights into the Photoactivated Ullmann Coupling Process Producing Highly Conjugated Oligothiophene Films on a Copper Substrate Sudarshan Natarajan,† Guangming Liu, and Seong H. Kim* Department of Chemical Engineering, PennsylVania State UniVersity, UniVersity Park, PennsylVania 16802 ReceiVed: January 20, 2006; In Final Form: March 3, 2006
This paper describes the details of surface reactions producing >100-nm-thick conjugated polymer films. When 2,5-diiodothiophene films deposited on copper are irradiated with UV at room temperature in Ar environments, oligothiophene films are synthesized. The average conjugation length of the produced film varies from ∼7 to 3-4 as the film thickness increases from ∼100 to ∼500 nm. The X-ray photoelectron spectroscopy analysis of the produced films reveals evidence for the formation of organo-copper intermediate species at the copper-monomer film interface and their diffusion from the copper surface into the monomer film during the photochemical process. A one-dimensional diffusion-reaction model is presented to explain the formation, diffusion, and reaction of organo-copper intermediates in the multilayer film during the photochemical reaction. The model simulation results qualitatively explain the decrease of the Ullmann coupling contribution in the photochemical reaction with the film thickness.
Introduction One of the necessary conditions for the continuous progress of a heterogeneous reaction at the solid surface is the formation of readily desorbing products. If the products do not desorb readily from the solid surface, then they act as a barrier for further access to the active surface sites, and this can stop further progress of the heterogeneous reaction. A well-known example in this regard is the formation of a passive oxide layer on aluminum. In this example, the formation of surface oxide layers works for our purpose because it protects the aluminum metal from further oxidation. In other heterogeneous reactions such as catalysis, the non-desorbing product formation can be problematic because it can stop the reaction after completion of the first monolayer reaction. A few successful examples of heterogeneous catalytic polymerizations are the Ziegler-Natta polymerization and surface-initiated atom-transfer radical polymerization.1,2 In the former example, the monomer dissolves in the polymer thin film and diffuses through it to reach the catalytic surface site. The monomer is then inserted between the catalytic site and the polymer. In the latter example, the polymer is anchored by a chemical bond at the substrate surface, and the polymerization site migrates to the end of the tethered polymer chain. When one attempts to apply organo-metallic reactions to the growth of conjugated polymer thin films and micropatterns directly on the substrate surface from the monomer, the formation of non-desorbing solid product poses a significant challenge. One of the most widely used methods for the synthesis of conjugated polymers is the Ullmann coupling reaction.3,4 In a typical Ullmann reaction, an aryl halide reactant is dissolved in a solvent and refluxed for several hours at high temperature (typically >100 °C) in the presence of Cu.3,4 In these conditions, the organo-copper intermediate species is formed at the copper surface and dissolved into the solution, * Author to whom correspondence should be addressed. E-mail: shkim@ engr.psu.edu. † Present address: Lam Research Corp., Fremont, CA 94538.
regenerating active copper sites; however, the polymer is formed in solution in an intractable powder form. In the absence of solvent, the Ullmann coupling reaction can also be carried out directly on active metal surface sites by adsorbing reactant molecules from the gas phase. In the 1990s, Bent and co-workers pioneered this thermal chemistry in cryogenic ultrahigh-vacuum (UHV) conditions.5-7 Gellman, Koel, and others also made significant contributions to the fundamental understanding of thermally activated Ullmann reactions.8-10 However, the thermal activation of the C-I bond can produce only sub-monolayerthick dimer molecules. Often, these dimer molecules desorb into the vacuum upon formation, and the surface is passivated with adsorbed iodine. This is because the desorption of iodinecontaining species into the gas phase is thermodynamically unfavorable at the reaction temperature. So, only the first monolayer molecules that are in direct contact with the substrate surface are thermally activated and participate in the desired reaction. There is no involvement of molecules in the second and higher layers of the adsorbed monomer film. To attain conjugated polymer films and micropatterns that are thick enough for technical applications (for example, 10100 nm), one must be able to continuously remove reaction products for the regeneration of active copper sites and supply reactants to the active copper sites in a spatially localized monomer film deposited on an active metal substrate. We recently demonstrated the feasibility of using photoactivated Ullmann coupling reactions for the production of ∼100-nmthick oligothiophene films and micropatterns in inert ambient conditions.11 In this process, a thick film of 2,5-diiodothiophene is deposited on the clean copper substrate and is activated photochemically with UV at room temperature. 2,5-Diiodothiophene is solid at room temperature (mp ) 34 °C), so it can be vapor-deposited by a simple evaporation method. The use of photoactivation avoids the need for thermal activation, so the monomer film can be kept adsorbed at the surface long enough for chemical reactions to occur. The average conjugation length of the oligothiophene film produced on copper is ∼7, which is long enough to mimic the electronic properties of
10.1021/jp0604239 CCC: $30.25 © 2006 American Chemical Society Published on Web 03/30/2006
8048 J. Phys. Chem. B, Vol. 110, No. 15, 2006 polythiophene. This conjugation length is significantly higher than the average conjugation length of 3-4 for oligothiophenes photochemically produced on inert substrates such as silicon oxide or gold.12-14 This paper describes the details of the process and the reaction pathways occurring at the copper and film interface as well as in the multilayer film. The thickness range of the film is from 100 to 500 nm. As the film thickness increases in this range, the photoluminescence color of the film changes from red to green. The conjugation-length distribution appears to be uniform throughout the entire film. These results are not simply due to the limited diffusion of the longer-conjugation-length oligothiophene species (produced at the copper surface) in the monomer film. The X-ray photoelectron spectroscopy analysis of the film and the diffusion-reaction model simulation of the process suggest that the thienyl-copper species, an intermediate of the Ullmann coupling reaction, diffuse and react in the deposited film. Experimental Section A thin copper foil (>99.99% purity) was used as the substrate. The copper surface was treated with 1.5 N nitric acid prior to use to remove oxide layers as well as organic contaminants.15 Contaminated copper substrates produced oligothiophene films with green photoluminescence emission, indicating no effect of the copper-substrate-mediated Ullmann reactions. 2,5-Diiodothiophene (monomer) was thermally evaporated at ∼120 °C in Ar ambient and deposited onto the copper substrate held at room temperature. The deposited monomer film was irradiated with a collimated UV beam from a 200 W mercury lamp (Oriel 6283) for 3 min in the argon environment. The IR radiation from the UV lamp was removed using a water filter. Tapping mode atomic force microscopy (AFM) measurements were used to measure the thickness of the photochemically produced oligothiophene films. On an average, the monomer deposition for 10 and 25 s under our experimental conditions yielded a final polymer film thickness of ∼100 and ∼250 nm, respectively. Thicker polymer films were made by increasing the monomer deposition time correspondingly. A solvent extraction method was used to check if there is any distribution of the conjugation length. The short-conjugation-length oligothiophenes are soluble in tetrahydrofuran (THF), but the longconjugation-length oligothiophenes are not. The photochemically synthesized oligothiophene films were analyzed with photoluminescence (PL) and X-ray photoelectron spectroscopy (XPS). The excitation source for PL spectroscopy was 435 ( 2 nm. The Ullmann reaction byproduct, CuI, does not have any PL emission at this excitation wavelength.11 A monochromatic Al KR source is used for XPS analyses, and the binding energies reported in this paper were referenced to C 1s at 284.6 eV. We have checked if the clean copper surface reacts with 2,5diiodothiophene liquid at ∼50 °C and 2-iodothiophene liquid at room temperature. Even after exposure of the clean copper substrate to these liquids for >10 min, there was no detectable photoluminescence. This simple experiment cannot rule out the monolayer-thick dimer or trimer formation, but it can clearly rule out that there is no significant Ullmann coupling reaction without photoactivation under our experimental conditions. Results and Discussions The Ullmann coupling contribution in the photochemical conversion of 2,5-diiodothiophene on the copper surface can be estimated from the average conjugation length of oligo-
Natarajan et al.
Figure 1. Photoluminescence emission spectra of oligothiophene films photochemically produced on copper with three different thicknesses: 100, 250, and 550 nm.
thiophene in the produced film. The PL emission spectrum is used for this purpose. Without Ullmann coupling reactions during the photochemical conversion, the conjugation length of the produced film is only 3-4, emitting greenish yellow PL.12-14 If Ullmann coupling reactions take place actively during the photochemical process, then the produced film has longer conjugation lengths and emits longer wavelength PL. Figure 1 compares the PL spectra of oligothiophene films of three different thicknesses photochemically produced on copper substrates. The PL emission maxima of the films of thicknesses of ∼100, ∼250, and ∼550 nm are found to be 610 nm (orange red), 575 nm (yellow), and 540 nm (greenish yellow), respectively. These spectra correspond to a conjugation length of ∼7, 4-5, and 3-4, respectively.16 Since all deposition and UV irradiation conditions are the same for these three samples, it is clear that the film thickness influences the extent of the Ullmann coupling contribution during the photochemical reactions in the 2,5-diioothiophene film deposited on copper. The thicker the film is, the shorter the conjugation length is. There is almost no sign of Ullmann coupling at the thickness of 550 nm as its PL emission is just the same as that of the oligothiophene film synthesized on nonreactive substrates such as silica and gold.12-14 Another evidence of the Ullmann coupling reaction during the photochemical reaction in the 2,5-diiodothiophene film deposited on copper can be found from the XPS analysis of copper iodide, a byproduct of the Ullmann coupling reaction. Figure 2 compares the I 3d5/2 XPS peaks for three different thicknesses of oligothiophene films (100, 250, and 550 nm) produced on copper. The peak at 620.5 eV is from the iodine covalently bonded to carbon, and the peak at 619.7 eV is from the iodide anion in CuI.17,18 As the film thickness increases, the CuI peak decreases at 619.7 eV, and the C-I peak increases at 620.5 eV. For the 550 nm film, the CuI portion almost completely disappears, and all of the detected iodine species is ascribed to the unreacted C-I bond of short-conjugation-length oligomers. In Figure 3, the Cu L3M4,5M4,5 Auger peaks are compared for three different thicknesses of oligothiophene films (100, 250, and 550 nm) produced on copper.19 Comparison with the reference spectra of CuI (KE ) 916.1 eV) and Cu (KE ) 918.5 eV) clearly shows that all of the copper species detected in XPS are CuI, not metallic Cu. There is no discernible peak at ∼917 eV corresponding to the Cu Auger peaks from CuS and Cu2S.20
Oligothiophene Films on a Cu Substrate
Figure 2. High-resolution XPS data of the I 3d5/2 region for oligothiophene films photochemically produced on copper with three different thicknesses: 100, 250, and 550 nm.
Figure 3. Cu L3M4,5M4,5 Auger peaks for oligothiophene films photochemically produced on copper with three different thicknesses: 100, 250, and 550 nm. For comparison, the reference spectra of CuI and Cu are also shown.21
This indicates that the desulfurization reaction of the thiophene ring with copper is negligible.11 The intensity of the Cu+ Auger peak decreases in proportion to the I- peak intensity at 619.17 eV. The Cu/I ratio is around 1, indicating the stoichiometry of CuI. It is found that the copper-to-carbon ratio drops by 80% and 98% when the film thickness is increased from 100 nm to 250 and 550 nm, respectively. It is important to note that the CuI species are detected for films much thicker than the photoelectron escape depth (typically 2-3 nm).21 This indicates that the CuI species are distributed throughout the entire film. If the CuI species are present underneath the 100-nm-thick and homogeneous oligothiophene film, then they will not be detected in XPS. The 100 nm thickness is much larger than the mean free path of the photoelectrons from the buried substrate. For the same thickness oligothiophene film produced on inert substrates, the substrate XPS peaks are not detectable at all.12 Therefore, the CuI detection in high-conjugation-length films thicker than 100 nm indicates that copper-containing species are desorbed from the substrate and diffused into the reacting film. The detection of CuI in the film provides a very important clue for understanding the reaction dynamics. CuI is not soluble in thiophene or 2-iodothiophene liquid. If the CuI species is produced only at the substrate/film interface, then it will be accumulated beneath the film and not be detected in XPS. In addition, the accumulation of CuI at the substrate/film interface will stop the Ullmann coupling reaction after the completion of one, or a few, monolayer-thick CuI layer formation. In a control experiment, the CuI film was intentionally formed on
J. Phys. Chem. B, Vol. 110, No. 15, 2006 8049 the Cu surface, and then the photochemical reaction of 2,5diiodothiophene was attempted, but there was no sign of the Ullmann coupling reaction producing films with red PL. Therefore, CuI cannot be the species desorbing from the copper surface and diffusing in the film during the photochemical reaction. Because CuI was ruled out, the diffusing species must be the thienyl-copper intermediates formed by the reaction of the photogenerated thienyl radical with the copper atom at the substrate surface. The reaction of the thienyl-copper intermediate with iodothiophene produces the C-C bond between the thiophene units. In this reaction, CuI is produced as a byproduct. One could speculate that the variation of the average conjugation length with the film thickness originates from the limited diffusion of the thienyl-copper intermediate species. If the intermediates can reach only up to 100-200 nm, then the film thicker than this limit would have a wide distribution of conjugation lengths. It would result in a higher conjugation length near the substrate/film interface (i.e., within 100 nm) and a lower conjugation length as the distance from the substrate increases. This possibility can be ruled out for the following reasons. If the thick film consisted of oligothiophenes with a conjugation length ranging from 3 (green PL) to 7 (red PL), then it would show a broader or even bimodal PL spectrum. But, regardless of the film thickness and the PL peak position, the width of the observed PL peak is almost the same. In the case of 250-nm-thick film, the PL spectrum shows the homogeneous emission of yellow wavelength light, instead of the mixed emission of green and red wavelength lights from different conjugation lengths. If the low-conjugation-length product is too thick on top of the low-conjugation-length product, then the PL excitation light might not reach the bottom layer, and the observed PL emission might be dominated by the low-conjugation-length product. This case can easily be checked with solvent fractionation. The low-conjugation-length oligothiophene is soluble in THF, but the high-conjugationlength one is not. When the 550-nm-thick film with green PL emission was dissolved in THF, we did not observe a film with red PL emission underneath. These results imply that the diffusion rate of the thienyl-copper intermediate species is at least comparable to, or even faster than, the reaction rate of the intermediate species with the iodothiophene species. It should be noted that when the 2,5-diiodothiophene monomer film is irradiated with UV light, a liquid phase is formed transiently in the monomer film before the entire film is solidified due to accumulation of oligothiophene molecules. This liquid phase is probably due to monoiodothiophene species formed via photodissociation of diiodothiophene. 2-Iodothiophene is liquid at room temperature. As in the conventional solution-phase process, the thienyl-copper species is expected to be soluble into this liquid phase. It is interesting to note that CuI is the only iodine-containing side product when the film thickness is ∼100 nm where the Ullmann coupling is dominating. The XPS peak position of I2 is very close to and almost indistinguishable from that of the covalently bonded iodine (620.5 eV). In thin films with a thickness of ∼100 nm, there is no discernible peak at 620.5 eV due to unreacted C-I species or the I2 side product. The photochemical reaction seems to be driven to the complete consumption of the covalently bonded iodine in the monomer molecule into CuI. This result implies that the desorption of the thienyl-copper species from the substrate and its diffusion and reaction to form CuI in the film are thermodynamically favorable. In this situation, the thickness dependence of the average conjugation length of oligothiophene molecules in the film must
8050 J. Phys. Chem. B, Vol. 110, No. 15, 2006
Natarajan et al. Cl,t/C*. Then, eq 1 is reduced to
∂2Cx,t' ∂x
Figure 4. Schematics of the photon flux distribution in the deposited 2,5-diiodothiophene film. The concentration of organo-copper intermediates depends on the intensity of UV photons reaching the surface of copper.
be controlled by the concentration of the thienyl-copper intermediate species produced at the copper/film interface. The concentration of the intermediate species will be determined by the UV photon intensity reaching the copper/film interface. Since UV photons are absorbed by the monomer molecules, the UV flux will decrease exponentially as it travels through the film (Figure 4). This can be modeled with the LambertBeer law. When the monomer film thickness is L, then the photon intensity at the copper surface (Io) can be expressed as Io ) IL × 10-ecL, where IL is the intensity of the photons at the surface of the monomer, c is the concentration of the monomer, and e is the extinction coefficient of the monomer. The concentration (c) of a solid film of pure monomer is 0.00253 mol/cm3. From UV-vis absorbance measurements of 2,5diiodothiophene in solution, the extinction coefficient (e) of the monomer is estimated to be 7.8 × 106 cm2/mol. This gives Io ) IL × 10-19734L (with L in centimeters). Therefore, the thienylcopper intermediate species formation will be less for a thicker film. With these insights, a qualitative diffusion-reaction model is developed to explain the extent of the Ullmann coupling contribution in the photochemical reaction of 2,5-diiodothiophene film on copper. For simplification, the model is constructed for one dimension as a function of the distance (l) from the substrate. We set l ) 0 at the copper/film interface and l ) L at the film/gas interface. The diffusion and reaction of the thienyl-copper intermediate species in the film can be modeled as22
D
∂2Cl,t ∂l
2
)
∂Cl,t + KCl,t ∂t
(1)
where Cl,t is the concentration of the thienyl-copper intermediate in the monomer film, D is the diffusion constant of the intermediate in the monomer film, and K is the rate constant for the reaction between the thienyl-copper intermediate and the monomer (or any oligomers containing C-I bonds at the end). Here, the monomer concentration is in excess compared to the intermediate concentration, so the K is a pseudo-firstorder rate constant. To solve this equation, steady-state boundary conditions are assumed such that C0,t ) C* (at the copper/film interface) and (∂Cl,t/∂l)L,t ) 0 (at the film/gas interface). The initial condition is Cl,t)0 ) 0 throughout the entire film. To simplify the numerical solution, we introduce the following dimensionless terms: x ) l/L, t′ ) Dt/L2, and Cx,t )
2
)
∂Cx,t′ + K'Cx,t′ ∂t′
(2)
where K′ ) KL2/D. The boundary and initial conditions of eq 2 are C0,t ) 1, (∂Cx,t/∂x)1,t′ ) 0, and Cx,0 ) 0. Assuming that all or a constant fraction of the photogenerated thienyl radicals react with the copper at the copper/film interface, the concentration of thienyl-copper species at the interface can be set to be directly proportional to the intensity of UV photons arriving at the copper surface: C* ∝ IL × 10-19734L. Therefore, the dimensionless concentration of the thienyl-copper intermediates at the copper/ film interface can be written as C0,t ) 10-19734L. As mentioned previously, there is a liquid phase formed at the beginning of the photochemical reaction. So, the diffusion of thienyl-copper intermediates during the photochemical reaction seems to occur through the liquid phase. In this numerical calculation, the diffusion constant of the thienyl-copper intermediate in the film (D) is assumed to be 10-6 cm2/s, which is at the lower end of typical diffusion constants in liquids.22 The diffusion constant will decrease significantly as the oligomers are formed and the film becomes solidified. Then, no more reaction can occur in the film. Since there is no quantitative data for diffusion coefficients of the intermediate species in the liquid film at different stages, we will limit this simulation only to the initial stage where D can be fixed to be constant. The rate constant (K) of the reaction between thienyl-copper species and monomers (or oligomers) needs to be estimated. If K is very high, then the oligomers will form immediately in the vicinity of the copper and prevent further diffusion of the thienyl-copper intermediates; hence Ullmann coupling reactions cannot occur in the monomer film. However, this is not the case at least in the 100-nm-thick film. So, we assume K to be on the order of D/L2 for L ) 100 nm. In that case, K′ for 100 nm becomes 1. The same K value is used for the simulation of other film thicknesses. Through the use of these parameters, eq 2 is numerically solved employing an implicit finite difference method.23 Once Cx,t is obtained, then the CuI concentration profile can be obtained from the following equation
[
]
d[CuI] dt'
x,t'
) K'Cx,t
(3)
The CuI concentration qualitatively represents the extent of the Ullmann coupling reaction contribution. The results of the numerical solution of eq 3 are plotted in Figure 5. In the figure, the concentrations of CuI accumulated at given positions (l) after irradiating the monomer film for 0.1 s are plotted for various monomer thicknesses. There is more CuI formed at the copper/monomer film interface for a thinner monomer film. As the monomer film thickness increases, the CuI amount produced at the copper surface decreases rapidly. The simulation results are qualitatively consistent with the experimental observations. Therefore, the Ullmann coupling contribution during the photochemical reaction decreases with the film thickness. If this profile is integrated over the entire period of irradiation, then one can obtain the overall contribution of the Ullmann coupling reaction throughout the entire UV irradiation period. But, as discussed before, the effective diffusion coefficient varies as the reaction progresses, so eq 3 cannot be solved for the entire UV irradiation period without knowing further details of the thienyl-copper diffusivity at different stages.
Oligothiophene Films on a Cu Substrate
J. Phys. Chem. B, Vol. 110, No. 15, 2006 8051 for thin film growth via surface-mediated organo-metallic reactions. In the case of highly conjugated oligothiophene synthesis from 2,5-diiodothiophene on copper, this limit is about 100-200 nm. In principle, this surface-mediated organo-metallic reaction can be applied to thin film fabrication of other systems as long as the reaction condition allows continuous regeneration of active sites in the deposited reactant film. Acknowledgment. This work is supported by the National Science Foundation (Grant No. DMI-0210229), the American Chemical Society Petroleum Research Foundation (Grant No. 40605-G5), and the 3M Nontenured Faculty Grant. References and Notes
Figure 5. Numerical simulation results of the one-dimensional diffusion-reaction model (eq 3) showing the effect of monomer film thickness (L) on the relative amount of CuI formed in the film after 0.1 s of irradiation. The diffusion coefficient (D) is assumed to be 10-6 cm2/s, and the rate constant K is assumed to be D/L2.
Conclusions The key for the successful growth >100-nm-thick films of the highly conjugated oligothiophene via surface-mediated Ullmann coupling reactions is the continuous regeneration of the active copper sites in the deposited monomer film. This requirement can be met if the intermediate species formed at the copper substrate surface desorbs and diffuses into the reacting film. The photoactivation of the condensed film of 2,5diiodothiophene on the clean copper surface at room temperature meets this requirement. A thin layer of transient liquid phase is formed upon UV irradiation, which allows the desorption and diffusion of thienyl-copper intermediate species from the substrate surface. However, this process cannot occur readily if the deposited monomer is too thick (i.e., >500 nm). This is because the UV photon flux reached at the copper/film interface is not sufficient enough to generate a high concentration of intermediate species. This photon flux decay can limit the maximum thickness of the deposited monomer that can be used
(1) Kim, S. H.; Somorjai, G. A. Catal. Lett. 2000, 68, 7. (2) Pintauer, T.; Matyjaszewski, K. Coord. Chem. ReV. 2005, 249, 1155. (3) Rieke, R. D. Science 1989, 246, 1260. (4) Fu¨rstner, A. Angew. Chem., Int. Ed. Engl. 1993, 32, 164. (5) Xi, M.; Bent, B. E. J. Am. Chem. Soc. 1993, 115, 7426. (6) Xi, M.; Bent, B. E. Langmuir 1994, 10, 505. (7) Kash, D. W.; Sun, D. H.; Xi, M.; Flynn, G. W.; Bent, B. E. J. Phys. Chem. 1996, 100, 16621. (8) Meyers, J. M.; Gellman, A. J. Surf. Sci. 1995, 337, 40. (9) Syomin, D.; Koel, B. E. Surf. Sci. 2001, 490, 265. (10) Hla, S.-W.; Bartels, L.; Meyer, G.; Rider, K.-H. Phys. ReV. Lett. 2000, 85, 2777. (11) Natarajan, S.; Kim, S. H. Chem. Commun. 2006, 729. (12) Natarajan, S.; Kim, S. H. Thin Solid Films 2006, 496, 606. (13) Natarajan, S.; Kim, S. H. Langmuir 2005, 21, 7052. (14) Liu, G.; Natarajan, S.; Kim, S. H. Surf. Sci. 2005, 592, L305. (15) Sung, M. M.; Sung, K.; Kim, C. G.; Lee, S. S.; Kim, Y. J. Phys. Chem B 2000, 104, 2237. (16) Fichou, D.; Ziegler, C. In Handbook of Oligothiophenes and Polythiophenes; Fichou, D., Ed.; Wiley-VCH: Weinheim, Germany, 1999. (17) Vasquez, R. P. Surf. Sci. Spectra 1994, 2, 149. (18) Sloan, D.; Sun, Y. M.; Ihm, H.; White, J. M. J. Phys. Chem. B 1998, 102, 6825. (19) Note that there is no difference in the Cu 2p peak positions for CuI and Cu. (20) Chawla, S. K.; Sankarraman, N.; Payer, J. H. J. Electron Spectrosc. Relat. Phenom. 1992, 61, 1. (21) Tanuma, S.; Powell, C. J.; Penn, D. R. Surf. Interface Anal. 1993, 21, 165. (22) Bird, R. B.; Stewart, W. E.; Lightfoot E. N. Transport Phenomena; Wiley: Singapore, 1994. (23) Ames, W. F. Numerical Methods for Partial Differential Equations; Academic Press: New York, 1977.
