pubs.acs.org/langmuir © 2009 American Chemical Society
Silica Nanoparticle Layer-by-Layer Assembly on Gold Feng Wang,† Scott Peters,‡ Jeff Guzda,‡ Richard H. Blunk,§ and Anastasios P. Angelopoulos†,* †
Department of Chemical and Materials Engineering, Univesity of Cincinnati, Cincinnati, Ohio 45227, ‡Fuel Cell Development Center, General Motors Corporation, Honeoye Falls, New York 14472 and §Fuel Cell Research Laboratory, General Motors Corporation, Warren, Michigan 48090 Received October 12, 2008. Revised Manuscript Received January 23, 2009
Layer-by-layer (LBL) assembly of silica nanoparticles is investigated as a means of controlling the surface wetting properties of gold electroplated onto 316 L stainless-steel substrates while maintaining a low electrical surface contact resistance. The strong polyelectrolyte acrylamide/β-methacryl-oxyethyl-trimethyl-ammonium copolymer is used as the cationic binder. The impact of silica nanoparticle zeta (ζ) potential for a range of -37.1 to 5.9 mV in the thickness, wettability, and contact resistance of the final LBL-assembled coatings is presented. The ζ potential is varied by altering both the pH and alcohol (ethanol) content of the silica suspensions and polymer suspension, consistent with the predictions of the Debye-Huckel equation. Nanoparticle adsorption is found to occur rapidly, with surface coverage equilibration obtained after only 1 min and uptake that is nearly linear with respect to the number of bilayers deposited. An increase in the absolute value of the (negative) ζ potential in the silica suspension is found to increase the bilayer thickness to an average value as high as 82% of the individual nanoparticle diameter for the smaller nanoparticles investigated, suggesting that nearly complete surface coverage may be achieved after the application of only a single nanoparticle-polymer bilayer (a coating thickness as low as 15.6 nm) and that nanoparticle adsorption is enhanced by electrostatic attraction between substrate and adsorbate. Counterintuitively, a more porous bilayer structure is observed if the ζ potential of the previously deposited nanoparticles is increased while the substrate is immersed in the cationic copolymer suspension, suggesting that copolymer adsorption in inhibited by substrate-solvent interactions. Wetting measurements demonstrate that silica LBL assembly results in a substantial reduction in contact angle from 84° on the bare substrate surface to as low as 15 ° after the application of a single bilayer and 7° after the application of eight bilayers. A monotonic increase in coating contact resistance is observed with an increase in the thickness with a characteristic volumetric electrical through-plane resistivity of as low as 1.63 kΩ 3 cm obtained from contact resistance measurement.
Corresponding author. E-mail:
[email protected].
after coating with a perfluorinated silane, mimics the superhydrophobic behavior of lotus leaves.6,7 The transparency and antireflection properties of such films for their use on windows and car windshields has been demonstrated.8 LBL assembly of oppositely charged silica and titania nanoparticles without intervening polyelectrolyte binder layers has also recently been shown to produce superhydrophilic coatings with antifogging properties.9,10 The pH dependence of the LBL assembly process for these films as well as films produced using silica and aminopropyl-functionalized silica nanoparticles11 has been well-characterized. In this investigation, silica nanoparticle LBL assembly on surfaces is evaluated as a method to control the surface hydrophilicity of electrically conducting substrates. For example, liquid-water transport through capillaries present in the otherwise electrically conductive components of protonexchange membrane (PEM) fuel cells can be particularly important for efficient system operation.12 Capillary-driven
(1) Iler, R. K. J. Am. Ceram. Soc. 1964, 47, 194–198. (2) Iler, R. K.Plural Monolayer Article and Process for Making Same. U. S. Patent 3,485,658,1969. (3) Hammond, P. T. Curr. Opin. Colloid Interface Sci. 2000, 4, 430–442. (4) Ariga, K.; Lvov, Y.; Ichinose, I.; Kunitake, T. Appl. Clay Sci. 1999, 15, 137–152. (5) Solberg, D.; Wagberg, L. Colloids Surf., A 2003, 219, 161–172. (6) Zhai, L.; Cebeci, F. C.; Cohen, R. E.; Rubner, M. F. Nano Lett. 2004, 4, 1349–1353. (7) Cheng, Y. T.; Rodak, D. E.; Angelopoulos, A. P.; Gacek, T. Appl. Phys. Lett. 2005, 87, 194112.
(8) Bravo, J.; Zhai, L.; Wu, Z.; Cohen, R. E.; Rubner, M. F. Langmuir 2007, 23, 7293–7298. (9) Lee, D.; Rubner, M. F.; Cohen, R. E. Nano Lett. 2006, 6, 2305–2312. (10) Lee, D.; Omolade, D.; Cohen, R. E.; Rubner, M. F. Chem. Mater. 2007, 19, 1427–1433. (11) Lee, D.; Gemici, X.; Rubner, M. F.; Cohen, R. E. Langmuir 2007, 23, 8833–8837. (12) Trabold, T. A.;Owejan, J. P.Flow Field Geometries for Improved Water Management. U.S. Patent 7,087,337, 2006.
I.
Introduction
Silica nanoparticle adsorption on various surfaces has been the subject of interest for many years as a means of developing a new class of nanostructured materials. Investigations were initiated by Iler in 1964,1,2 who deposited alternating layers of silica and alumina nanoparticles as a means of varying the degree of dispersion and thixotropic behavior of ceramic slips. Silica layer-by-layer (LBL) assembly has been proposed as a means of creating ordered arrays for photonic bandgap applications,3 and studies of various silica assemblies have appeared.4 The adsorption of single silica nanoparticle layers onto dispersed cellulose fibers has been explored as a means of enhancing fiber retention in papermaking.5 A more recent application that has been investigated is the use of silica nanoparticle LBL assembly to create a textured surface that, *
4384
DOI: 10.1021/la8033758
Published on Web 3/9/2009
Langmuir 2009, 25(8), 4384–4392
Wang et al.
instabilities have been shown to result in liquid-water holdup.13 In the case of PEM fuel cells, such a holdup presents a barrier to the diffusion of reactant gases, which can significantly limit the available current density. Hydrophilic coatings applied to the surfaces of bipolar plate channels in PEM fuel cells have been shown to mitigate this problem.14 The work presented in this article demonstrates that LBL assembly permits the application of hydrophilic coatings that are sufficiently thin to minimize the impact to the contact resistance between electrically conductive components. Another key advantage of LBL assembly relative to more typical coating methods (e.g., spray, roll, and dip) is that lowmolecular-weight species that assist in maintaining silica suspension stability (e.g., various ionic compounds) may be rinsed away during processing, leaving behind only the tightly adhered silica nanoparticles on the substrate surface. Consequently, PEM system contamination concerns are minimized when liquid flows over a bipolar plate channel coating prepared by utilizing LBL assembly. We have recently demonstrated that polyelectrolyte binders can yield very low contact resistance in LBL-assembled coatings prepared from electrically conductive but relatively hydrophobic nanoparticles.15 Consequently, multilayer films produced by LBL assembly of silica nanoparticles with such binders are expected to possess substantially lower contact resistance than the alternate approach which employs oppositely charged silica and titania nanoparticles 8,9 while retaining hydrophilic character. Film hydrophilicity arises due to the negative surface charge which results from the dissociation of surface silanol groups on the hydrolyzed silica surface.16 The detailed kinetics and physicochemical interactions responsible for silica multilayer formation with cationic polyelectrolytes will be investigated in this paper using two commercial silica dispersions. A number of aspects of silica LBL assembly are unique to the present investigation: 1) The effect of suspension alcohol content in concert with other suspension parameters on the electrostatic potential near the surface of the nanoparticles, as quantitatively assessed by the zeta (ζ) potential measurement, and the consequent impact on LBL assembly kinetics with polyelectrolyte are systematically evaluated. Suspension parameters are chosen on the basis of the Debye-Huckel equation and thus provide a mechanistic interpretation of LBL assembly behavior. There exist previous published investigations of silica LBL assembly that have examined the effect of suspension pH and associated changes to ζ potential on equilibrated all-nanoparticle structures.10,11 The effect of salt addition (ionic strength) on the adsorption of single silica layers onto cellulose fibers using polyelectrolyte has also been evaluated.5 Otherwise, studies have focused primarily on demonstrating the feasibility of silica LBL assembly for various applications 1,2,4-6,8,9 using particular multilayer structures. 2) Substrates consist of gold electroplated onto 316 L stainless steel, a potential bipolar material for PEM fuel cells. More typical in LBL studies is the use of substrates such as (13) Son, S. Y.; Allen, J. S.Visualization and Predictive Modeling of TwoPhase Flow Regime Transition with Application Towards Water Management in the Gas-Flow Channels of PEM Fuel Cells. Proceeding of IMECE2005, Nov. 5-11, 2005, Orlando, Florida. (14) Angelopoulos, A. P.; Peters, S.Nanoparticle Coating Process for Fuel Cell Components. U.S. Patent Application 20,070,141,2382007. (15) Alazemmi, , M.; Wang,, F.; Blunk, , R.; Angelopoulos, , A. P. Adv. Funct. Mater. , accepted for publication, 2009. (16) Iler, R. K. The Chemistry of Silica; Wiley: New York, 1979.
Langmuir 2009, 25(8), 4384–4392
Article
silica.1,5,6,8-11 In addition, silver electrodes have been used in LBL studies when a quartz crystal microbalance (QCM) was used to monitor silica weight uptake due to multilayer adsorption.4 However, to our knowledge, silica adsorption on gold has not been previously investigated. 3) Contact resistance measurements are employed to assess the volumetric through-plane resistivity. Such evaluations are critical in assessing the impact of electrically insulating multilayer coatings such as silica on substrates employed in electronics applications and have never been previously performed. 4) We demonstrate how Monte-Carlo simulation of the electron scattering and X-ray emission events in energydispersive spectroscopy (EDS) measurements may be used to obtain LBL thickness. EDS is a more widely accessible research tool than either QCM4 or spectroscopic ellipsometry.5,8-11 In this work, ζ potential measurement is used to quantitatively assess the extent of the electrostatic interaction between nanoparticles in a given suspending medium and subsequent to surface deposition. ζ potential is the electrostatic potential at beginning of the diffuse double layer near the nanoparticle surface (including adsorbed ions) and determines the mobile counterion distribution which extends into the bulk of the suspending media. The relationship between ζ potential and the nanoparticle surface charge, σ, and properties of the suspending media such as dielectric constant, ε, and ionic strength (as reflected in the Debye-Huckel parameter, κ), is given by the Gouy-Chapman Theory:17 -2KkTε zeζ sinh σ ¼ ze 2kT
ð1Þ
where k is Boltzmann’s constant, T is the absolute temperature, e is the electron charge, and only a symmetric electrolyte is considered such that the charge z = z+ = -z-. From this equation, we may predict how suspending media composition may be altered to systematically modify multilayer nanostructure: (1) Debye parameter, κ, may be varied by indifferent salt addition as well as the nanoparticle concentration, (2) dielectric constant, ε, may be varied through the addition of alcohols, and (3) surface charge, σ, may be altered by varying the nanoparticle or polyelectrolyte synthesis procedure or, where surface charge results from the dissociation of weakly acidic silanol groups, the pH of the suspending media. The influence of nanoparticle ζ potential on LBL adsorption behavior will be systematically examined in the present investigation. The influence of suspension parameters on ζ potential may be predicted more simply than suggested by Eq. (1) by noting that the Debye parameter, κ, for all suspensions investigated in this study has a value >0.5 nm -1 and the measured particle radius, a, has a value at or greater than ∼20 nm. Consequently, κa > 10 and the double layer is thin relative to the particle radius and Eq. (1) may be shown to simplify to the Debye-Huckel approximation:17 ζ ¼
-σ εK
ð2Þ
Thus, ζ is inversely proportional to κ and ε and directly proportional to σ. (17) Hunter, R. J. Foundations of Colloid Science; Oxford University Press: New York, 1992; Vol. 1.
DOI: 10.1021/la8033758
4385
Article
Wang et al.
II.
Experimental Section
A. Materials. i. Adsorbates. Two types of silica nanoparticles of differing sizes were obtained for this study from commercial sources. Polishing silica (from Electron Microscopy Supply, hereafter referred to as EMS) was obtained in an alkaline (NH4OH) base at 40% solids content with a nominal particle size of 60 nm from Electron Microscopy Sciences. The second commercial silica obtained is known as X-TEC 3408 (hereafter referred to as NanoX) and is available from NanoX Corporation. The material comes in a 10% by volume ethanol in water base at a solids content of 5% by weight and has a nominal particle size of 20 nm. The cationic binder that will be employed in this study is a copolymer of acrylamide and a quaternary ammonium salt (acrylamide/β-methacryl-oxyethyl-trimethyl-ammonium copolymer). The specific material employed has a charge density of 10% and is available commercially as Superfloc C442 from Cytec Corporation. This is a strong polyelectrolyte with a cationic charge density which is independent of pH. The cationic binder has been described in detail in previous investigations.15,18,19 ζ potential measurements as a function of pH have also been previously presented for this material.15 To summarize: a positive ζ potential was obtained at all pH values investigated, with the magnitude increasing from a value of 7 mV at pH 1 to a maximum value of 29 mV at pH 5. As the pH increases from 5 to 7, a decrease in ζ potential is observed. Such a decrease is consistent with the onset of carboxylic acid group dissociation and the presence of amphiphilic groups along the polymer chain.18 ii. Substrates. Electroplated gold of 200 nm thickness on 316 L stainless was employed as the bipolar plate substrate material for silica multilayer formation. Irreversible adsorption of the cationic polyacrylamide to be employed in this study has been previously observed on various substrate surfaces, including electroplated gold.19,20 As will be described shortly, use of gold as the substrate material also permits accurate assessment of the electrical resistivity of the deposited layers. Substrates were cut into 1 in. x 1 in. squares and then wiped with methanol (Fischer Scientific, ACS Spectrophotometric grade) and a lint-free cloth to remove any fingerprints prior to coating application. B. Suspension Formulation and Multilayer Formation.
Nanoparticle ζ potential was manipulated first by maintaining the pH in the silica colloid suspensions constant and varying the dielectric constant. The pH of both the anionic silica suspension and the corresponding cationic binder was varied using H2SO4 (ACS Plus grade, Fisher Scientific) to maintain the desired pH. Ethanol (Fischer Scientific, ACS Spectrophotometric grade) was employed to vary the dielectric constant. A pH value of 5.0 was chosen for the silica suspensions since this value is well away from the region where the hydrolyzed acrylamide groups on the cationic binder ionize and yet far enough above the silica iso-electric point (IEP) to easily maintain suspension stability. The impact of surface charge was next investigated by varying the pH of the cationic polyelectrolyte suspension above and below the IEP of the previously adsorbed silica nanoparticles. This approach was preferred over varying the pH of the silica suspensions themselves, since, as will be described in detail shortly, nanoparticles were found to agglomerate at suspension pH values near or below the IEP. NanoX concentration was kept at 10% by volume in suspension (a solids content of 0.5%) to both minimize use of raw material and permit evaluation of an alcohol content (18) Angelopoulos, A. P.; Benziger, J. B.; Wesson, S. P. J. Colloid Interface Sci. 1997, 185, 147–156. (19) Angelopoulos, A. P.; Matienzo, L. J.; Benziger, J. B. J. Colloid Interface Sci. 1999, 212, 419–425. (20) Angelopoulos, A. P.; Cangelosi, J.; Kotylo, J. A.; Matienzo, L. J.; Shaver, N. L.Promoting Adhesion between a Polymer and a Metallic Substrate. U.S. Patent 6,908,684, 2005.
4386
DOI: 10.1021/la8033758
approaching 1% by volume. In the case of the EMS silica, the concentration chosen was 5% by volume (a solids content of 2%). The effect of silica concentration was not systematically investigated in the present work but a quantitative impact was only observed with a more than 2-fold increase in concentration. Only deionized (DI) water was employed for dilution and rinsing in this work. DI water was obtained by passing a 1 MΩ resistivity house water supply through a Millipore Synthesis water purification unit to achieve 18.2 MΩ. LBL deposition occurred first with application of the polymer (for 2 min) followed by thorough rinsing (1 min total in two baths with vigorous agitation) then immersion into the silica suspension for the desired time interval, followed again by thorough rinsing (1 min total in two baths with vigorous agitation). This deposition cycle was repeated as many times as desired. The chemical baths remained quiescent during immersion of the samples but were thoroughly agitated between sample immersions. Polymer immersion time was chosen based on the known rapid adsorption of this material.18-20 C. Characterization Methods. Zeta potential measurements were obtained at a temperature of 25 °C with the aid of photon correlation spectroscopy (PCS, Zeta-Plus from Brookhaven Instruments). Electrolyte concentrations were ∼0.05 M for all suspensions employed in the present study. Thus, κa > 10 and the double layer is thin relative to the particle radius. Consequently, ζ potential was calculated using the Smoluchowski approximation.17 Determination of the ζ potential via the Smoluchowski method requires information concerning the dielectric constant and viscosity of the solvent. The values used are shown in Table 1. Dielectric constants were obtained experimentally using a dielectric constant meter from Brookhaven Instruments (model BI-870) and are compared to literature values 22 in Table 1. The refractive indexes employed for light scattering measurements are also shown in Table 1.21 PCS was also used to assess nanoparticle diffusivity and, assuming an equivalent hydrodynamic sphere and applying the Stokes-Einstein equation, the diameter of the nanoparticle. The as-prepared nanoparticle suspensions were found to be sufficiently dilute for stable photon correlation coefficients to be obtained, obviating the need for further dilution. FE-SEM (FEI XL-30 System), was initially used in this investigation in an attempt to obtain high resolution images the multilayers deposited. Unfortunately, it was not possible to differentiate the silica nanoparticles from the texture of the electroplated gold surface utilizing this technique due to the insulating properties of the silica (substantial charging of the surface occurred). Application of FE-SEM (Zeiss Ultra 55) with an in-lens detector 24 and low accelerating potentials (1 keV) was subsequently found capable of imaging the silica nanoparticles and employed for select coatings to confirm the PCS nanoparticle size measurement technique. A typical image obtained utilizing this technique is shown in the Table of Contents figure, which depicts the surface of 8 layers of 20 nm NanoX silica nanoparticles deposited onto electroplated gold via LBL assembly with cationic polyacrylamide at 100 kX magnification. EDS (EDAX Corporation), was employed to monitor changes in the elemental composition of the coating as a function of LBL deposition and suspension immersion time. Elemental analysis is useful in these systems due to the unique composition of the silica nanoparticles relative to the substrate (i.e., Si deposited on Au). EDS detects X-ray photons emitted from the scattering volume of incident electrons. Since the (21) Scott, T. A. J. Phys. Chem. 1946, 50, 406–412. (22) Franks, F., Water: A Comprehensive Treatise. . Plenum Press: , 1973; Vol. 2. (23) Wensink, E.; Hoffmann, A. C.; Maaren, P.; Spoel, D. J. Chem. Phys. 2003, 119, 7308–7317. (24) Goldstein, J. Scanning Electron Microscopy and X-ray Microanalysis, 3rd ed.; Kluwer Academic/Plenum Publishers: New York, 2003.
Langmuir 2009, 25(8), 4384–4392
Wang et al.
Article
Table 1. Effect of Ethanol Content in Aqueous Solution ethanol volume % properties
0%
20%
50%
refractive index (from ref 21) dielectric constant (measured) dielectric constant (from ref 22) viscosity (cp) (from ref 23)
1.33 80.8 80.4 0.89
1.35 71.2 71.1 1.82
1.36 55.4 54.1 2.40
scattering volume penetrates a few microns into the substrate surface, EDS samples through the entire thickness of the multilayers. Semiquantitative analysis was performed at a constant accelerating potential of 15 keV, a working distance of 10 mm, an amp time of 51.2 μs and a magnification of 200X. The Si K emission intensity was compared to the Au L emission intensity for relative analysis with results reported as weight % Si relative to Au, C, and O (i.e., the elements of the underlying stainless steel were excluded from the analysis). A minimum of 3 measurements on each of the two sides of the substrates were taken with the average and standard deviation of all points reported. The thickness for a given LBL assembled coating was obtained from EDS data using Monte-Carlo electron flight simulations (CASINO Version 2.42, Universite de Sherbrook, Quebec) under the same microscope conditions as those used experimentally.24 Dynamic contact angle measurements were made utilizing a Sigma 701 wetting balance (KSV Instruments). The probe liquid was DI water and the stage was moved at a velocity of 5 mm/ min, ensuring quasi equilibrium advancing and receding contact angles. A minimum of 3 readings were performed per sample. At least 3 back-to-back advancing-receding cycles were performed per sample to ensure that the contact angle did not change with repeated immersions (suggesting loss of material or absorption of water, either of these invalidating the contact angle result). Electrical contact resistance was assessed using a custom built hand-press from Carver. Gold coated platens were attached to swivel points on the press and connected to a power supply. Teflon tape on one platen surface defined a 1 in2 contact area into which was placed a 1 in2 coated sample. The sample was first sandwiched between two conductive compliant sheets of graphite fiber paper (Toray TGP-H-1.0T) to permit the controlled application of a contact pressure of 200 psi for all measurements. The through-plane potential drop at an applied current density of 1 A/cm2 (6.5 Amps) was read which, when subtracted by the potential drop for the bare substrate and divided by two (the sample being coated on both sides), gave the coating contact resistance, Rc, in units of Ω-cm2. In other words: Rc = RA where, R is the through plane resistance of the coating and A is the contact area perpendicular to the direction of current flow. The through plane volume electrical resistivity, F, of the coating may be obtained from this contact resistance and the estimated thickness, t, of the coating using: F ¼
Rc t
ð3Þ
where the coating surface is assumed to conform perfectly to the electrode surfaces. Since this is unlikely in the case of LBLassembled nanoparticle coatings, Equation 3 provides only an upper bound on the volume resistivity in the present study.
III.
Results and Discussion
A. Silica Nanoparticle ζ Potential and Size. Silica suspensions were prepared as described in the Experimental section of the paper with a solids content of 0.5% and 2% by weight for the NanoX and EMS silica, respectively. These suspension concentrations were sufficiently dilute to permit the direct Langmuir 2009, 25(8), 4384–4392
Figure 1. ζ potential of NanoX nanoparticles in a 0.5% solids suspension at the indicated pH and ethanol volume % in water.
Figure 2. ζ potential of EMS nanoparticles in a 2% solids suspension at the indicated pH and ethanol volume % in water. application of PCS for ζ potential and particle size analysis (i.e., stable correlation coefficients could be obtained). ζ potential measurements for NanoX and EMS are shown in Figures 1 and 2, respectively, as a function of ethanol content and the pH of the suspending media. The label of “1%” ethanol in these and subsequent figures for NanoX represents a suspension prepared without added ethanol. In the case of NanoX, ethanol is present in the raw material (10% by volume) and results in the presence of a small amount of alcohol even in the highly diluted suspension. Each point in the figure represents the average of three measurements. Error bars represent ( one standard deviation from the mean. In general, we note that an increase in the alcohol content in the suspensions results in an increase in the absolute value of the ζ potential. The increase in ζ potential is consistent with the decrease in suspension dielectric constant as shown in Table 1. This increase is particularly evident at pH values greater than 4 in Figures 1 and 2. The IEP for both types of silica occurs at a pH value of about 2 and is independent of the alcohol content. The presence of ethanol therefore does not appear to interfere with the extent of silanol dissociation (e.g., through surface adsorption 17). Particle size measurement for the silica nanoparticles were performed using PCS as a function of pH in regions where the suspension remained stable (pH values above the IEP). Results for NanoX are shown in Table 2. The average particle diameter from all pH values in Table 3 is 19.0 nm DOI: 10.1021/la8033758
4387
Article
Wang et al.
Table 2. Number-Average Particle Diameter versus pH for the NanoX Suspension pH
size (nm)
standard deviation (nm)
polydispersity
2.96 4.07 5.02 6.08
18.7 19.4 18.7 19.1
0.6 0.4 0.7 0.6
0.28 0.29 0.27 0.31
Table 3. Number-Average Particle Diameter versus pH for EMS Suspension pH
size (nm)
standard deviation (nm)
polydispersity
3.08 4.01 5.07 6.00
66 67 62 66
1 2 1 2
0.11 0.12 0.12 0.14
with a standard deviation of 0.3 nm. Polydispersity was found to be quite high in these samples, with an average value of 0.29. Standard deviations and polydispersities at each pH value are provided in the Table. Results for the EMS nanoparticles are shown in Table 3. The average particle diameter is 65 nm with a standard deviation of 2 nm. As in the case for NanoX, the average polydispersity was found to be rather high at 0.12. B. Multilayer Formation with the Cationic Polymer at pH 1 (below the IEP of Silica Nanoparticles). Equilibration of nanoparticle coverage was found to occur rapidly for both commercial silica nanoparticles investigated. The impact of immersion time in 10% by volume NanoX silica suspension is shown by the EDS data in Figure 3. The ethanol content in the suspension was 50% by volume and the pH was 5. The polymer was at pH 1 and a concentration of 0.72 g/l. Consequently, the ζ potential of the nanoparticles was -37.1 mV in the silica suspension and 5.0 mV in the polymer suspension. Figure 3 depicts the weight % Si as a function of the number of bilayers (polymer-nanoparticle layers) deposited. Error bars in this figure represent ( one standard deviation from the average of a total of 3 measurements on each of two sides of the sample. To obtain the thickness data corresponding to Figure 3, Monte-Carlo simulations were performed as previously described in the Experimental section. The solid curve in Figure 4 depicts the results from such simulations where the k-ratio (the silicon, Si, X-ray intensity, I, from a silica, SiO2, coating on pure gold relative to that from a pure silicon reference, I ref) is plotted versus the coating thickness. We note that the plot is linear up to a thickness of at least 1000 nm. Consequently, the following equation may be used to determine the coating thickness, t, from an experimental determination of the k-ratio (I/Iref): t ¼ 1508 nm
I Iref
ð4Þ
where 1508 nm is the slope of the line in Figure 2 and represents the extrapolated coating thickness at which the k-ratio becomes 1. Figure 5 depicts the coating thicknesses obtained by applying Equation 4 to the experimental k ratio data corresponding to Figure 3. We note from both Figures 3 and 5 that substantial silica uptake occurs despite negligible electrostatic attraction between the cationic polymer and the 4388
DOI: 10.1021/la8033758
Figure 3. Si uptake (as measured via EDS) as a function of the number of bilayers deposited from a 10% by volume NanoX suspension at pH 5 and 50% ethanol in water by volume. The immersion time is as indicated. The cationic polymer is at 0.72 g/L and pH 1. The dashed line is for 2 min.
Figure 4. Monte Carlo simulations for the Si (L) k ratio (I/Iref) as a function of the thickness of a SiO2 coating.
Figure 5. Coating thickness corresponding to that in Figure 4. previously deposited silica nanoparticles. No silica deposition is observed in the absence of the cationic polymer and the small amount of Si detected by EDS on the bare substrate (∼0.5% by weight) results from the underlying stainless steel formulation. One possible source of attraction is hydrogen bonding between the acrylamide groups on the polymer and the weakly acidic silanol groups. Cationic polyacrylamide binding to basic functional groups through nonelectrostatic association has been previously demonstrated.15,18,19 Figures 3 and 5 show that there is little difference in terms of silica loading as a function of time for 1 and 2 min Langmuir 2009, 25(8), 4384–4392
Wang et al.
immersions, indicating the equilibration time-scale of surface adsorption. A substantial drop in nanoparticle deposition is observed at an immersion time of 10 s. Uptake is found to be approximately linear with the number bilayers deposited, regardless of immersion time. The bilayer thickness obtained from a linear fit after surface coverage equilibration (2 min.) is found to be 15.6 nm with a correlation coefficient of about 1.0. This value is not too different from the nanoparticle diameter obtained using PCS (Table 2, 19.0 nm) and suggests that the substrate surface is almost completely covered by silica after the application of only a single bilayer. To evaluate the impact of nanoparticle ζ potential while in suspension, the alcohol content of the suspension was decreased from 50% ethanol by volume to 20% and finally 1% (no added ethanol). Immersion time in the silica suspension was maintained at 2 min to ensure equilibration of surface coverage. Thickness results for the silica coating obtained by applying Equation 4 to experimental k-ratio data is shown in Figure 6. We note that reduction in ζ potential yields a substantial reduction in thickness, with an equilibrated average bilayer thickness in the case of 1% suspension of 5.4 nm (one-third that obtained with 50% ethanol). Electrostatic interaction between the charged nanoparticles in the silica suspension at pH 5 and the deposited cationic polymer is thus also a critical component in multilayer formation in addition to immersion time. Changes to multilayer thickness due to alteration of nanoparticle suspension conditions may be predicted qualitatively on the basis of ζ potential measurement. This result is entirely consistent with previous investigations of silica LBL assembly using titania.10 An example of the appearance of silica nanoparticles deposited onto electroplated gold is shown in Table of Contents image, obtained utilizing the Zeiss Ultra 55 SEM. The coating in this image consists of 8 layers of NanoX silica nanoparticles deposited via LBL assembly with cationic polyacrylamide. Nanoparticle size in this image is comparable to that measured using PCS. The ζ potential of the EMS nanoparticles is -36.7 mV in the pH 5 silica suspension and 5.9 mV in the pH 1 polymer suspension. Thus, as in the case of NanoX, little electrostatic attraction is expected between the polymer in suspension and the nanoparticle surface. Adsorption equilibration time is found to be comparable for both types of silicas (∼1 min). The primary difference in utilizing EMS silica instead of
Figure 6. Coating thickness as a function of the number of bilayers deposited from a 10% by volume NanoX suspension at pH 5 and % ethanol in water by volume as indicated in the plot. The immersion time is 2 min in every case. The cationic polymer is at 0.72 g/L and pH 1. Langmuir 2009, 25(8), 4384–4392
Article
NanoX is the substantially greater Si uptake which results from the larger size of the EMS nanoparticles. In addition, the bilayer thickness begins at a value below that of the diameter of a single nanoparticle and decreases with each application, even at the immersion time required for adsorption equilibration. Such behavior may arise from enhanced cumulative repulsion between successive applications of EMS nanoparticles and the associated substrate surface morphology (although the ζ potentials of the individual NanoX and EMS nanoparticles are comparable). Bilayer thickness equilibration occurs after 4 bilayers have been deposited. As with NanoX, the thickest bilayer is obtained with 50% ethanol and, after 4 bilayers have been deposited, and has a value of 42.2 nm. This value is substantially lower than the EMS nanoparticle thickness (Table 3, 65.2 nm) and suggests that the equilibrated EMS bilayer is more porous (65% estimated surface coverage) than that of the NanoX bilayer (82% estimated surface coverage). C. Multilayer Formation with the Cationic Polymer at pH 5 (above the IEP of the Silica Nanoparticles). The impact of increasing the absolute value of the (negative) surface charge of previously deposited silica nanoparticles while in polymer was next investigated. The kinetic studies discussed in the previous section were repeated with the only change being an increase in the pH of the polymer suspension to a value of 5. This value is well above the IEP of both types of nanoparticles and results in a ζ potential of -23.7 mV for NanoX and -27.3 mV for EMS while in the polymer. Results for NanoX thickness are shown in Figure 7, with similar qualitative results obtained for EMS. The most surprising change in behavior relative to the pH 1 polymer solution is that Si uptake and corresponding thickness are substantially reduced. Maximum equilibrated bilayer thickness is 7.4 nm for NanoX and 22.3 nm for EMS. These values represent a reduction in loading of 50% relative to the pH 1 polymer data. This change in uptake with ζ potential is opposite that previously described with alcohol content and is counter-intuitive. Such behavior is not consistent with the standard Scheutjens-Fleer description of polymer adsorption 25 in that an increase of the Debye thickness, κ-1, and the associated increase in the substrate and adsorbate ζ potentials is expected to enhance polymer adsorption. Analogous behavior has been observed in a previous study of the effect of ionic strength on cationic polyacrylamide adsorption onto a previously deposited silica nanoparticle surface.5 We have also previously encountered such behavior of cationic polyacrylamide adsorption on charged 26 and uncharged 19 surfaces. A model has been developed 27 to account for such behavior which recognizes that adsorption energy is attenuated by the presence of solvation energy (i.e., the attraction of water molecules for both substrate and adsorbate). Thus, as substrate-adsorbate attraction increases, so does attraction to the solvent and a point is reached where an adsorption maximum is reached, whereupon uptake paradoxically declines with increasing interaction energy. Such a model would be consistent with the observations in the present investigation associated with increasing polymer pH and the absolute value of the (negative) ζ potential. (25) Van de Steeg, H. G. M.; Cohen Stuart, M. A.; de Keizer, A.; Bijterbosch, B. H. Langmuir 1992, 8, 2538–2546. (26) Angelopoulos, A. P.; Kim, Y. Fuel 2002, 81, 2167–2171. (27) Angelopoulos, A. P. J. Colloid Interface Sci. 2001, 243, 292–299.
DOI: 10.1021/la8033758
4389
Article
Wang et al.
Figure 7. Coating thickness as a function of the number of bilayers deposited from a 10% by volume NanoX suspension at pH 5 and 50% ethanol in water by volume. The immersion time is as indicated. The cationic polymer is at 0.72 g/L and pH 5. The dashed line is for 2 min. Interestingly, a maximum in silica deposition with increasing ζ potential was observed in the recent study of silica nanoparticle LBL assembly with oppositely charged aminopropyl-functionalized silica nanoparticles.11 One possible explanation provided in that investigation was that increased silica nanoparticle ζ potential enhanced electrostatic repulsion between intervening layers and thus reduced nanoparticle deposition, presumably while the aminopropylfunctionalized silica particles remained at a fixed bilayer thickness. Separation of the contribution to the coating thickness by each of the two oppositely charged nanoparticles would have strengthened this hypothesis. In addition, one must account for the contrasting behavior observed when titania is used.10 Regardless, the applicability of ζ potential correlations to adsorption behavior in all-nanoparticle versus nanoparticle-polyelectrolyte LBL systems is uncertain and more research in the field (both experimental and theoretical) is required. D. Silica Multilayer Surface Energetics. The impact of silica suspension alcohol content on the advancing contact angle of a NanoX and EMS coating is depicted by Figures 8 and 9 for equilibrated bilayer thicknesses (2 min immersion in silica suspension). For NanoX, an increase in the average thickness of the coating with increased alcohol content is observed to make the surface more hydrophilic. The most hydrophilic surface in Figure 8 has a contact angle of 17 ° after 8 bilayers from 50% ethanol suspension versus a contact angle of 84 ° for the bare substrate. Substantial wetting hysteresis was observed. The reduction in advancing contact angle observed with increasing silica uptake may be explained by an increase in adsorbate (silica) surface coverage as described by the Johnson and Dettre model for contact angle hysteresis.28 Interestingly, despite the substantially lower average bilayer thickness, the wetting behavior of the NanoX coating prepared from pH 5 polymer is comparable to that prepared from pH 1 polymer (50% ethanol in the silica suspension for both cases) and is among the most hydrophilic of all samples. According to the Johnson and Dettre model, such wetting behavior would indicate comparable silica surface area coverage for the two sets of samples. If that is the case, then the pH of the polymer suspension is not only impacting the porosity of each bilayer, but also the extent of bilayer interpenetration during LBL assembly. Such a hypothesis 4390
DOI: 10.1021/la8033758
Figure 8. Advancing contact angle of NanoX coatings corresponding to Figures 3, 5, 6, and 7. The key indicates the % by volume ethanol-in-silica suspension and the pH of the cationic polymer suspension.
Figure 9. Advancing contact angle of EMS coatings. The key indicates the % by volume ethanol-in-silica suspension and the pH of the cationic polymer suspension. is not unreasonable in light of the fact the extent of polymer adsorption onto previously deposited silica nanoparticles will be mediated by a change in polymer suspension pH and associated nanoparticle ζ potential. The surface wetting behavior of LBL-assembled coatings is somewhat different in the case of EMS (Figure 9) in that the advancing contact angle appears to be relatively insensitive to changes in alcohol content of the silica suspension from which they were obtained. Such behavior is consistent with constant silica surface coverage and a high degree of bilayer interpenetration. As in the case of NanoX, wetting hysteresis was observed. The lowest advancing contact angle observed in the case of EMS coatings is 10 ° after 8 bilayer from 50% ethanol suspension. E. Silica Multilayer Contact Resistance. Contact resistance values for equilibrated (2 min immersion in silica suspension) NanoX coatings are plotted in Figure 10. The contact resistance monotonically increases with the number of bilayers deposited and qualitatively corresponds to Si uptake. Additional comparison is made by dividing the data Langmuir 2009, 25(8), 4384–4392
Wang et al.
Article
Figure 10. Contact resistance, Rc, of NanoX coatings corresponding to those in Figures 3, 5, 6, and 7. The key indicates the % by volume ethanol-in-silica suspension and the pH of the cationic polymer suspension.
Figure 12. Contact resistance, Rc, of the EMS coatings. The key indicates the % by volume ethanol-in-silica suspension and the pH of the cationic polymer suspension.1
Figure 11. Volumetric electrical through-plane resistivity associated with Figure 10.
in Figure 10 by corresponding values of the coating thickness (Figures 3 and 5) to obtain the volumetric through-plane resistivity according to Equation 3. The result is shown in Figure 11. We note that only in the case of the pH 5 polymer does the resistivity appear to plateau to a characteristic value of 1.63 kΩ 3 cm. This value is within the range required for the electrostatic discharge of electronic components.29 This value is also about half of that obtained from a coating deposited with a polymer suspension at pH 1 and a silica suspension containing 1% ethanol (3.17 kΩ 3 cm at eight bilayers), which has the same average thickness. However, the same average thickness between the two coatings is associated with enhanced hydrophilicity (higher silica surface coverage) in the case of the coating prepared from a pH 5 polymer suspension. This observation suggests that the coating prepared from the pH 5 polymer is more densely packed and actually thinner, perhaps yielding greater opportunity for direct electrical contact between the two conducting plates. However, an analysis of all coatings subsequent to contact resistance testing did not reveal any damage within the resolution of our SEM instrumentation. The characteristic volumetric electrical resistivity of all coatings is substantially less than the electrical resistivity of fused silica (∼1015 kΩ 3 cm). The volumetric resistivity for the other coatings is just observed to begin to plateau in Figure 11. Such behavior suggests that a uniform coating structure has not been achieved even after the application of eight bilayers. The maximum resistivity observed for coatings prepared from pH 1 polymer is 5.85 kΩ 3 cm. Contact resistance values for the EMS nanoparticle coatings are shown in Figure 12. Values were found to be an order of magnitude higher than for NanoX coatings with a similar number of bilayers. Volumetric resistivities scale to roughly Langmuir 2009, 25(8), 4384–4392
twice that of the NanoX coatings. EMS coatings prepared from the pH 1 polymer exhibit characteristic volumetric resistivities with an upper bound ranging from 12.4 to 18.2 kΩ 3 cm. Coatings prepared with EMS silica are thus insulating materials.29 As with the NanoX coatings, EMS coatings deposited with the polymer suspension at pH 5 have a contact resistance and volumetric resistivity substantially less than coatings with a similar average thickness prepared from a pH 1 polymer suspension, again suggesting a higher nanoparticle packing density for coatings prepared from a pH 5 polymer.
IV.
Conclusions
A silica nanoparticle LBL assembly was investigated as a unique method of controlling the surface wetting properties of metallic substrates employed as bipolar plates in PEM fuel cells while maintaining a low electrical surface contact resistance. Nanoparticle adsorption occurred rapidly, with an equilibration of surface coverage observed after only 1 min and a nearly linear increase in coating thickness with respect to the number of bilayers deposited. An increase in the absolute value of the (negative) ζ potential in the NanoX silica suspension was found to raise the bilayer thickness to an average value of as high as 82% of the individual nanoparticle diameter. Such behavior suggested that nearly complete surface coverage may be achieved after the application of only a single nanoparticle-polymer bilayer (a coating thickness as low as 15.6 nm) and that nanoparticle adsorption was enhanced by electrostatic attraction between substrate and adsorbate. A more porous bilayer structure was observed with larger silica EMS nanoparticles with a maximum bilayer thickness of 42.2 nm or 65% of the nanoparticle thickness. Counterintuitively, a more porous bilayer structure was observed if the ζ potential of the previously deposited nanoparticles was increased while immersed in the cationic copolymer suspension. The implication of such behavior was that copolymer adsorption is inhibited by substrate-solvent interactions. LBL assembly yielded a substantial reduction in the contact angle from 84° on the bare substrate surface to as low as 15° after the application of a single bilayer and 7° after the application of eight bilayers. DOI: 10.1021/la8033758
4391
Article
Wang et al.
A monotonic increase in coating contact resistance was observed with an increase in thickness with a characteristic volumetric resistivity of as low as 1.63 kΩ 3 cm with NanoX silica. This value was within that required for numerous conductive coating applications and substantially less than that for fused silica.
4392
DOI: 10.1021/la8033758
Coatings prepared from the LBL assembly of EMS silica nanoparticles have been characterized as insulators. Acknowledgment. Funding for this work has been provided by the General Motors Corporation and is gratefully acknowledged.
Langmuir 2009, 25(8), 4384–4392
