Polyelectrolyte Adsorption on Thin Cellulose Films Studied with

Thin cellulose films were prepared by dissolving carboxymethylated cellulose fibers in N-methyl morpholine oxide and forming thin films on silicon waf...
0 downloads 0 Views 483KB Size
134

Biomacromolecules 2009, 10, 134–141

Polyelectrolyte Adsorption on Thin Cellulose Films Studied with Reflectometry and Quartz Crystal Microgravimetry with Dissipation Lars-Erik Enarsson and Lars Wa˚gberg* Department of Fibre and Polymer Technology, The Royal Institute of Technology, Teknikringen 56, 100 44 Stockholm, Sweden Received September 11, 2008; Revised Manuscript Received October 24, 2008

Thin cellulose films were prepared by dissolving carboxymethylated cellulose fibers in N-methyl morpholine oxide and forming thin films on silicon wafers by spin-coating. The adsorption of cationic polyacrylamides and polydiallyldimethylammonium chloride onto these films was studied by stagnation point adsorption reflectometry (SPAR) and by quartz crystal microgravimetry with dissipation (QCM-D). The polyelectrolyte adsorption was studied by SPAR as a function of salt concentration, and it was found that the adsorption maximum was located at 1 mM NaCl for polyelectrolytes of low charge density and at 10 mM NaCl for polyelectrolytes of high charge density. Electrostatic screening led to complete elimination of the polyelectrolyte adsorption at salt concentrations of 300 mM NaCl. According to the QCM-D analysis, the cellulose films showed a pronounced swelling in water that took several hours to complete. Subsequent adsorption of polyelectrolytes onto the cellulose films led to a release of water from the cellulose, an effect that was substantial for polyelectrolytes of high charge density at low salt concentrations. The total mass change including water could therefore show either an increase or a decrease during adsorption onto the cellulose films, depending on the experimental conditions.

Introduction The interaction between cellulose and other compounds is of interest in many applications since it controls, for example, the adhesion, wetting and absorption properties of cellulosebased materials. Cellulose frequently occurs in, for example, wood-based composite materials, cotton textiles, viscose fibers, soluble cellulose derivatives, papers, and absorbents.1 Cellulose fibers are however complicated to use in studies of cellulose interactions since they constitute an inhomogeneous, microscopic material with rough, fibrillated surfaces that often exclude direct surface analysis techniques. In this respect, fundamental knowledge of cellulose interactions has greatly benefited from the development of smooth cellulose films with a thickness in the range from a few nm to 100 nm. Such films can be used as a model system for cellulose fibers, and this has made it possible to study molecular interactions between test substances and cellulosic materials with new techniques in situ. This was exemplified early by Buchholz et al.2 who studied the adsorption of cationic polyacrylamide on cellulose using the surface plasmon resonance technique. A thorough review of the preparation and use of model cellulose films is given by Kontturi et al.3 The films are typically prepared from a homogeneous solution of dissolved cellulose that is distributed over a smooth substrate followed by regeneration of the solid cellulose film. The first generation of cellulose model films was developed by Schaub et al.,4 who prepared the soluble derivative trimethylsilylcellulose (TMSC) and transferred this derivative to a hydrophobic substrate using the Langmuir-Blodgett technique and followed this by regeneration of the cellulose. This method ¨ sterberg and co-workers.5,6 has been further developed by O 7 Gunnars et al. reported an alternative method where the cellulose was dissolved in N-methylmorpholine oxide and spun * To whom correspondence should be addressed. Fax: +46 8 790 8101. E-mail: [email protected].

into a thin film on the substrate by means of spin-coating. This method gives films of cellulose II with a thickness of about 30 nm. The spin-coating technique is fast and has also been successfully applied with TMSC solutions8 and with cellulose dissolved in lithium chloride-dimethylacetamide.9,10 It has later been shown that cellulose model surfaces can also be prepared from fine dispersions of colloidal cellulose samples. Edgar and Gray11 degraded microcrystalline cellulose in strong acid to obtain cellulose nanocrystals that were used in the preparation of model surfaces with the spin-coating technique. Wågberg et al.12 and later Aulin et al.13 have used microfibrillated cellulose to make a model surface by consecutively adsorbing cationic polyelectrolytes and cellulosic nanofibrils according to the polyelectrolyte multilayer technique.14 Although much effort has been spent on the preparation and characterization of the cellulose films, there are relatively few studies that have quantified the adsorption of polyelectrolytes onto cellulose films.2,6,15 Such data are of interest in the field of fiber engineering that aims to develop new products from cellulose materials. It also brings new possibilities to the field of paper chemistry, where interaction between polyelectrolytes and cellulose is essential to enhance the fixation of the colloidal materials to the fiber surfaces and to improve the dry and wet strength of the paper produced from these fibers. The purpose of this study was to utilize the potential of the cellulose model surfaces and to study the molecular events that take place on the cellulose surface when polyelectrolytes are added to the system. Here, a combination of the stagnation point adsorption reflectometry (SPAR) and quartz crystal microgravimetry with dissipation (QCM-D) techniques was applied to identify the different contributions of the polyelectrolytes, water, and cellulose in the adsorption process. This gave information about the adsorption kinetics, swelling effects of the cellulose film and the structure of the adsorbed polyelectrolyte layers. Three polyelectrolytes with different charge

10.1021/bm8010193 CCC: $40.75  2009 American Chemical Society Published on Web 11/24/2008

Polyelectrolyte Adsorption on Thin Cellulose Films

Biomacromolecules, Vol. 10, No. 1, 2009

135

densities were studied and their adsorption was quantified as a function of the background electrolyte concentration. To simulate the charged nature of pulp fibers more closely, the cellulose material was carboxymethylated to different degrees, corresponding to one medium and one high level of the total charge density of conventional pulp fibers.

Experimental Section Materials. The cellulose source was a bleached sulfite dissolving pulp, Domsjo¨ Dissolving Plus, from Domsjo¨ Fabriker, Domsjo¨, Sweden. Cationic polyacrylamides (CPAM) were provided by Professor Hiroo Tanaka, Faculty of Agriculture, Kyoshi University, Fukuoka, Japan. These were copolymers of acrylamide and acrylamide-propyl-trimethylammonium chloride (APTAC) that were specially synthesized at a low degree of monomer conversion to obtain a polyelectrolyte with a narrow distribution in charge density of the individual chains. The two samples included in this study were denoted CPAM1 and CPAM4 and their charge densities were 0.50 and 2.4 meq/g, respectively. The weightaverage molecular masses were determined by size exclusion chromatography to be 4.6 × 106 Da for CPAM1 and 3.8 × 106 Da for CPAM4, and the corresponding polydispersity indexes (Mw/Mn) were 4.06 and 5.05, respectively, according to a previously reported chromatography method.16 The charge densities were determined by polyelectrolyte titration17 against potassium polyvinyl sulfate (Wako Pure Chemical Industries, Osaka, Japan) using an instrumental colorimetric end point detection.18 A homopolymer of diallyldimethylammonium chloride (pDADMAC) of commercial grade (Alcofix 109) was supplied by CDM Chemicals (Go¨teborg, Sweden). This polyelectrolyte had a theoretical charge density of 6.19 meq/g. The sample was ultrafiltered with a membrane cutoff of 500000 Da to isolate the high molecular weight fraction of the sample. The water used for preparing sample solutions and for washing the substrates was of Milli-Q ultrapure grade with a resistivity of 18.2 MΩ · cm. Silicon wafers of boron-doped, p-type (MEMC Electronic materials SpA, Novara, Italy) were oxidized for different times at 1000 °C to obtain top oxide layers of silica with thicknesses of 39 and 90 nm. Strips 10 × 50 mm in size were cut to fit the reflectometer cell. The oxide layer was determined using null ellipsometry in air at ambient conditions (Rudolph Research ellipsometer model 43702-200E, NJ, U.S.A.). Carboxymethylation of Cellulose Fibers. The dissolving pulp was modified with 1-chloro-acetic acid at two substitution levels, following the method of Walecka.19 The pulps were analyzed for total charge density according to the method of Katz,20 yielding 64 and 128 µeq/g, respectively, while the native dissolving pulp had a total charge density of 30 µeq/g. The increments corresponded to substitution degrees of 0.0055 and 0.015, respectively, and the yields of the carboxymethylation reaction were 82 and 94%, respectively. Preparation of Cellulose Films. Carboxymethylated dissolving pulp (0.25 g) was dissolved in 12.5 g N-methyl morpholine oxide (NMMO, 50%) under stirring at 125 °C. After complete dissolution, the viscous solution was diluted with 37.5 g dimethyl sulfoxide (DMSO) to give a cellulose concentration of 0.5% The diluted solution was kept thermostatted at 125 °C. The surface substrates, oxidized silicon wafers and QCM quartz crystals, were cleaned in 10 W plasma (model PDC002, Harrick Scientific Corp.,Ossining, NY, U.S.A.) and thereafter immersed in 0.1 g/L polyvinylamine solution at pH 8 for at least 15 min to form an adsorbed layer of polyvinylamine, which served as an anchoring layer for the cellulose. Immediately before film preparation, the substrates were rinsed in Milli-Q water and blown dry in a stream of nitrogen. Cellulose films were prepared using a spin-coater (model KW-4A-2, Chemat Technology, Northridge, CA, U.S.A.). A total of 100 µL of dissolved cellulose was applied on the substrate followed by spinning at 1500 rpm for 10 s and finally 3500 rpm for 30 s. Solid cellulose films were precipitated by immersing the substrate in Milli-Q

Figure 1. Overview of parameters in the five-layer optical model that describes the reflection interface at the surface. This set has been used for calculation of the sensitivity factor As-1, relating the adsorbed amount to the reflectivity at the surface. The light source had an incidence angle of 70°.

water. DMSO/NMMO residues in the film were removed by dialysis against Milli-Q water overnight. The substrates were then dried in a nitrogen stream and thereafter annealed at 105 °C for 6 h. After preparation, the model surfaces were stored in a desiccator until use in an adsorption experiment. Thickness Determination of Cellulose Films. The cellulose film thicknesses were analyzed by making a scratch in the cellulose film, according to the method of Fa¨lt et al.21 The resulting height difference between the cellulose layer and the base wafer was determined by atomic force microscopy operated in tapping mode. The instrumentation consisted of a Picoforce SPM equipped with a EV scanner (Veeco Instruments, Santa Barbara, CA, USA). Standard tapping-mode silicon cantilevers (RTESP) were supplied from the same source. The 95% confidence interval for the mean film thickness was 30 ( 5 nm. Stagnation Point Adsorption Reflectometry (SPAR). An instrument from the Laboratory of Physical Chemistry and Colloid Science, Wageningen University, Wageningen, The Netherlands, was used. A description of this technique has been given by Dijt et al.22 and the use of cellulose substrates with SPAR has been reported separately by Torn et al.23 Reflected light from the substrate is split into its parallel and perpendicular components and the intensity of each component is measured. The ratio, S, of these intensities constitutes the primary reflectometer output, which is usually separated into an initial baseline level, S0, and a shift due to adsorption, ∆S. For a thin film in the order of 10 nm, the latter is proportional to the adsorbed amount

Γ ) As-1 ·

∆S S0

(1)

where As-1 is a proportionality factor that can be determined by applying an optical model based on Abeles matrixes. These calculations were performed in the software “Reflec” supplied with the SPAR. In brief, the solid-liquid interface is here treated as a set of optical layers located between the two bulk phases, that is, silicon and water. Each optical layer is assumed to be homogeneous and defined by its refractive index and thickness. Figure 1 illustrates the optical layers present on a model cellulose surface and the applied parameters, adapted from Torn et al.,23 except for the thickness of the wet cellulose film that was estimated from the data of Fa¨lt et al.21 and the refractive index of the polyelectrolyte layer, which was estimated from Sennerfors et al.24 A parallel calculation of the refractive index of wet cellulose at different water contents is shown below in Figure 4. The refractive index increments of the CPAM1 and CPAM4 polyelectrolytes were determined to 0.174 and 0.173 mL/g, respectively, at a salt concentration of 0.1 M NaCl, using an Abbe Refractometer (Carl Zeiss, Oberkochen, Germany). The refractive index increment of pDADMAC, 0.176 mL/ g, was taken from Lingstro¨m et al.,25 as determined in both deionized water and 0.5 M NaCl. Quartz Crystal Microgravimetry with Dissipation (QCM-D). A D300 model from Q-Sense AB, Va¨stra Fro¨lunda, Sweden was used.

136

Biomacromolecules, Vol. 10, No. 1, 2009

Enarsson and Wågberg

Figure 2. Adsorption of 50 mg/L CPAM1 on carboxymethylated cellulose, DS ) 0.015, as a function of time with different oxide layer thicknesses on the supporting silicon wafers. Oxide layer thickness: triangles: 0 nm; circles: 39 nm; squares: 90 nm.

Figure 3. Sensitivity factors plotted as a function of the wet cellulose layer thickness. This is repeated for three different oxide layer thicknesses on the supporting silicon wafer. Oxide layer thickness: triangles, 0 nm; circles, 39 nm; squares, 90 nm. The vertical line indicates the assumed wet thickness of the prepared cellulose films.

Crystals with a top layer of sputtered silica were delivered by the same manufacturer. Data of the third overtone was used unless otherwise stated. The presented data are reported in normalized frequencies, i.e. the overtone frequencies are divided by the overtone number. The Sauerbrey relationship26 was used for the calculation of the adsorbed mass, ∆m

∆m ) C ·

∆f n

(2)

where ∆f is the resonance frequency, n is the overtone number, and C is a proportionality constant equal to -0.177 mg · m-2 · Hz-1, as derived by Edvardsson et al. for this type of crystal.27 The energy dissipation, D, from the oscillating crystal was also obtained with the QCM-D technique, according to the relationship28

D ) (πfτ)-1

(3)

where f is the fundamental resonance frequency and τ is the decay time constant for the oscillation amplitude. The dissipation constitutes a qualitative measure of the rigidity of the film.

Results Polyelectrolyte Adsorption on Cellulose Studied with Reflectometry. When using the SPAR for adsorption measurements on cellulose films, the first task was to determine its sensitivity and to optimize the method. This was done by adjusting the oxide layer thickness of the supporting wafer. Figure 2 shows three studies of the adsorption of CPAM1 on cellulose films carboxymethylated to DS ) 0.015, where the thickness of the oxide layer on the wafer was varied between 0 and 90 nm. A thickness of about 100 nm is recommended to optimize the sensitivity for adsorption studies on silica substrates,22 although 100 nm thick cellulose films have previously been studied by reflectometry on wafers with a native oxide layer of only 2-3 nm.23 The present data show that stable adsorption signals were obtained for all the oxide layer thicknesses studied but that the sensitivity of the SPAR method, as expected, was strongly dependent on the absolute thickness. The low drift in the signal during baseline equilibration indicated that the SPAR was practically insensitive to any swelling of the cellulose layer that occurred during wetting. In all cases, a finite adsorption plateau was reached upon saturation, even though the results with 0 nm oxide layer thickness showed a sudden shift in reflectivity during the rinsing step.

Figure 4. Refractive indices of wet cellulose (open squares) and corresponding sensitivity factors in SPAR (filled circles) calculated as a function of water uptake within the cellulose film. The dry thickness of the cellulose film was set to 30 nm.

The sensitivity of SPAR was also evaluated on a theoretical basis using the five-layer optical model, shown in Figure 1. This approach was specifically used to calculate the influence of the oxide layer thickness, as well as the effect of the thickness of the cellulose film and its water content. Results are presented in the form of sensitivity factors, which are used to convert the SPAR data into adsorbed amounts in mg/m2. Figure 3 shows the effect of a change in thickness of the cellulose film on the sensitivity factor, evaluated at three different oxide layer thicknesses. The calculations were performed for a wet cellulose film with a constant refractive index of 1.46, that is, at a constant relative water content. The vertical line at 60 nm represents the assumed wet cellulose thickness of the present cellulose films, based on earlier reported data for this type of carboxymethylated cellulose films.21 In agreement with the experimental observations in Figure 2, the model showed that the oxide layer thickness had a strong influence on the SPAR sensitivity. It also showed that sensitivity factor will increase with increasing thickness of the cellulose film, given that the relative water content is constant. A good way of testing the five-layer model was to apply the calculated sensitivity factors on the replicate measurements in Figure 2. If the model is representative, it should compensate for the differences in the oxide layer thickness and provide equal adsorbed amounts. The current model results showed adsorbed amounts of 6.8, 7.6, and 5.4

Polyelectrolyte Adsorption on Thin Cellulose Films

Figure 5. Saturation adsorption for the three polyelectrolytes studied with SPAR as a function of sodium chloride concentration at pH 7. Films of carboxymethylated cellulose with DS ) 0.015. Polyelectrolytes: triangles, 50 mg/L CPAM1; circles, 50 mg/L CPAM4; diamonds, 10 mg/L pDADMAC.

mg/m2 on wafers with an oxide layer thickness of 0, 39, and 90 nm, respectively. These results showed that the strong dependence of the primary adsorption signals on the oxide layer thickness was essentially compensated for by the derived optical model, which gave support for the applicability of the model. Another important theoretical test was to determine how the state of swelling of the cellulose film affected the sensitivity of SPAR. Figure 4 shows a model calculation of the refractive index of the wet cellulose film and the corresponding sensitivity factor, evaluated as a function of the water content in the layer. The refractive index of wet cellulose was determined according to the model of Holmberg et al.,5 using a linear combination of the refractive indices of dry cellulose and water, weighted against their volume fractions. The model clearly demonstrated that the refractive index of the cellulose film will decrease as it takes up water, while the sensitivity factor of the SPAR will remain nearly constant, showing a small increase. As an example, the sensitivity factor was predicted to increase by 4% when the water content in the film increases from 15 to 30 mg/ m2. This result gave theoretical support to the experimental observation that the SPAR signal was relatively insensitive to swelling effects during the baseline equilibration. Having established the sensitivity relationships for different optical thicknesses of the cellulose model surfaces in SPAR, the technique was applied to measure the adsorption of three different polyelectrolytes on cellulose films carboxymethylated to DS ) 0.015. A constant oxide layer thickness of 39 nm was used in all cases in order to facilitate a comparison of the results. Figure 5 shows the influence of NaCl addition on the saturation adsorption of two CPAMs and a pDADMAC sample, presented as ∆S/S0 from the SPAR measurements. The results are proportional to the adsorbed amount, since all the polyelectrolytes had refractive index increments of approximately 0.17 mL/ g. The highest adsorbed amount was found for the CPAM1 sample with a weak maximum at 1 mM NaCl. Above this salt concentration, the adsorption decreased at 10 mM and was insignificant at 100 mM NaCl, which clearly indicated that the adsorption of CPAM1 was very sensitive to electrostatic screening. The polyelectrolytes with higher charge densities, CPAM4 and pDADMAC, were adsorbed in lower amounts but they were, on the other hand, less sensitive to electrostatic screening, as their adsorption was still significant at 100 mM but became insignificant at 300 mM NaCl. The adsorption maximum of these polyelectrolytes was found at 10 mM, which

Biomacromolecules, Vol. 10, No. 1, 2009

137

indicated that the maximum was shifted upward with increasing polyelectrolyte charge density. In the case of pDADMAC, the adsorption at 100 mM was still greater than that at 0.1 and 1 mM NaCl. This illustrated that electrostatic screening at moderate salt concentrations enhanced the adsorption of highly charged polyelectrolytes. Taking into consideration the sensitivity factors in Figure 3, the adsorbed amounts in 0.1 mM NaCl were calculated to be 6.7 mg/m2 for CPAM1, 1.6 mg/m2 for CPAM4, and 0.7 mg/m2 for pDADMAC. Polyelectrolyte Adsorption on Cellulose Studied with QCM-D. When the QCM-D technique was used, additional information was obtained about the water included in the cellulose films and in the adsorbed layers. Typical results are shown in Figure 6, where the adsorption of CPAM1 onto a cellulose film with DS ) 0.015 is depicted in terms of the responses in frequency and dissipation. A long swelling phase took place after exposure of the dry cellulose film to salt solution and this process was followed until the film reached a state of almost constant frequency and dissipation. All data in the figure are related to measurements on the wet cellulose film, whereas the total frequency shift between the dry state and the wet swollen state after 20 h was ∆f3/3 ) -760 Hz (third overtone after normalization). At this point, indicated by zero time in the figure, the remaining drift in the normalized frequency was -0.005 Hz · min-1 and the drift in dissipation was +4 × 10-10 dissipation units · min-1, as evaluated for the third overtone. The three normalized overtones 3, 5, and 7 are plotted together to give a qualitative indication of the rigidity of the film. The strong similarities between the frequency shifts of different overtones suggested that the combined film of CPAM1 and cellulose had a rigid structure. The negative shifts indicated that the polyelectrolyte adsorption led to a net increase in mass on the model surface, about 6 mg/m2 when the Sauerbrey model was applied. The dissipation shifts were moderate, about 3 × 10-6 units, while the separation of the overtones here indicated a weak viscous effect of the CPAM1 layer. A second example of the results obtained with QCM-D is given in Figure 7 for the adsorption of pDADMAC in Milli-Q water on a cellulose film with a lower degree of carboxymethylation. This highly charged polyelectrolyte gave in the absence of a background electrolyte a positive frequency shift, +3 Hz in normalized frequency units, while the minor instant response in dissipation was negative, -0.1 × 10-6 units. Although the effects were small in magnitude, both signals indicated an apparent decrease in the mass sensed with QCMD. This was in contrast to the results obtained with SPAR that clearly showed an effective adsorption of pDADMAC on similar model films having a higher degree of carboxymethylation. The decrease in mass observed with QCM-D is therefore probably related to the water content in the film, suggesting that water was released from the cellulose film when the highly charged polyelectrolyte was adsorbed. To understand better the responses of polyelectrolyte adsorption observed with QCM-D, a systematic study was conducted that included the effects of polyelectrolyte charge density, the degree of carboxymethylation of the cellulose and the salt concentration. Figure 8 shows the results obtained for CPAM1 and CPAM4 adsorption on cellulose films with the lower degree of carboxymethylation. The graphs depict the initial shifts in frequency and dissipation that were obtained for saturation adsorption. Any long-term effects were disregarded since these mainly related to slow baseline drifts. CPAM1 showed generally negative frequency shifts, which indicated a net mass increase upon adsorption, except in 100 mM NaCl where no adsorption

138

Biomacromolecules, Vol. 10, No. 1, 2009

Enarsson and Wågberg

Figure 6. QCM-D data on the swelling of a carboxymethylated cellulose film with DS ) 0.015 in 1 mM NaCl at pH 7, followed by addition at time t ) 0 of 50 mg/L CPAM1 at the same salt concentration. (A) Normalized frequency shifts for three overtones, order from top: n ) 7, 5, and 3. (B) Dissipation shifts for three overtones, order from top: n ) 3, 5, and 7. The measurements were started immediately after the addition of salt solution and the transition from air to water is therefore not shown in the graphs.

Figure 7. QCM-D data for the swelling of a carboxymethylated cellulose film with DS ) 0.0055 in Milli-Q water, followed by adsorption of 50 mg/l pDADMAC dissolved in Milli-Q water. (A) Frequency shift and (B) dissipation shift. Data based on the seventh overtone. The insets shows a magnification of the adsorption during the first 150 min.

Figure 8. Total shifts at saturation adsorption of CPAM1 and CPAM4 on carboxymethylated cellulose with DS ) 0.0055 studied by QCM-D. Data from the third overtone showing (A) initial frequency shifts and (B) initital dissipation shifts. All measurements were made at pH 7 and 50 mg/L polyelectrolyte concentration.

was found. This agreed well with the effect of salt on the adsorbed amount that was determined by SPAR. The dissipation shifts increased with salt concentration up to 10 mM, which indicated a looser attachment of the polyelectrolyte to the surface with increasing salt concentration, likely due to a higher fraction of segments distributed as extended loops and tails. No response in dissipation was observed at 100 mM NaCl, which was taken

as a further indication that polyelectrolyte adsorption is totally screened at this salt concentration. The frequency response upon adsorption of CPAM4 with the higher charge density was generally lower in magnitude than that for CPAM1. At 0.1 mM salt the frequency increased upon adsorption, indicating a net mass decrease. Since a significant adsorption of CPAM4 was found with SPAR, the response in

Polyelectrolyte Adsorption on Thin Cellulose Films

Biomacromolecules, Vol. 10, No. 1, 2009

139

Figure 9. Total shifts at saturation adsorption of CPAM1 and CPAM4 on carboxymethylated cellulose with DS ) 0.015 studied with QCM-D. Data from the third overtone showing (A) initial frequency shifts and (B) initital dissipation shifts. All experiments were conducted at pH 7 and 50 mg/L polyelectrolyte concentration.

frequency indicated that the released mass of water dominated over the adsorbed amount, similar to the results found for pDADMC adsorption in deionized water. At 10 mM NaCl, the adsorption seemed to increase, since the frequency shift was clearly negative and the dissipation reached a local maximum. No signs of adsorption were seen in 100 mM NaCl either in the frequency or the dissipation. Figure 9 shows the effect of a higher degree of carboxymethylation of the cellulose on the adsorption of polyelectrolytes. In the case of CPAM1, the adsorption was clearly enhanced by an increase in the surface charge density, since both the frequency and dissipation shifts were larger than those shown in Figure 8. The results for CPAM4 adsorption showed a frequency shift that was positive in 0.1 mM NaCl but clearly negative in both 10 and 100 mM NaCl. A similar trend was seen in the dissipation that changed sign when the salt concentration was raised above 0.1 mM NaCl. The decrease in total moisture content dominated over the mass increase due to polyelectrolyte adsorption at the lowest salt concentration, while the situation was the opposite at higher salt concentrations. It was also observed that the combination of a high charge density of the polyelectrolyte and of the cellulose film resulted in an effective adsorption in 100 mM NaCl that was apparent in both QCM-D and SPAR data. This showed that the limiting salt concentration for total screening of the polyelectrolyte adsorption was shifted to higher concentrations when the charge densities of the polyelectrolyte and the substrate were simultaneously increased.

Discussion Adsorption of Polyelectrolytes onto Carboxymethylated Cellulose Studied by SPAR. The main effects of salt concentration and charge density on polyelectrolyte adsorption could be directly read from the original reflectometer output, because the oxide layer thickness was constant and the refractive index increments were similar for all the polyelectrolytes. These data indicate that the polyelectrolyte adsorption on cellulose was governed mainly by electrostatic interactions, because the adsorption was so sensitive to electrostatic screening at 100 mM. The salt effects also fitted well to the predictions of the Scheutjens-Fleer theory for polyelectrolyte adsorption.30 The results suggest that weakly charged polyelectrolytes have a high adsorption at low salt concentrations but are relatively sensitive to electrostatic screening, in contrast to highly charged poly-

electrolytes that adsorb in lower amounts but are more resistant to electrostatic screening. The conclusion that electrostatics dominates the adsorption energy on cellulose was also in agreement with previous studies on cellulose fibers, which indicate that nonionic interactions have an insignificant influence on the polyelectrolyte adsorption.31-33 To illustrate the effect of nonionic interactions, the total screening of CPAM1 adsorption at 100 mM can be compared with the adsorption data for a similar cationic polyacrylamide on silicon oxide, which showed a significant adsorption in the presence of over 300 mM salt.34,35 Polyelectrolyte adsorption on cellulose has been described as an ion exchange process at low salt concentrations.36 This condition was tested for the current results by calculating the adsorbed amounts and the corresponding number of adsorbed polyelectrolyte charges at 0.1 mM NaCl. Because all the polyelectrolytes had similar refractive index increments, the sensitivity factors presented in Figure 2 were also applicable to CPAM4 and pDADMAC. The corresponding numbers of adsorbed charges were calculated to be 3.4 µeq/m2 for CPAM1, 3.9 µeq/m2 for CPAM4, and 4.2 µeq/m2 for pDADMAC. The results indicated that the adsorption mechanism was close to a charge neutralization mechanism and that the equivalent surface charge was about 4 µeq/m2 for the carboxymethylated cellulose film with DS ) 0.015. This was slightly less than the calculated total charge density of the cellulose film, 5.8 µeq/m2, assuming a dry film thickness of 30 nm and dry density of 1500 kg/m3. The latter calculation indirectly suggested, as expected, that most of the cellulose charges interacted with adsorbed polyelectrolytes. Adsorption of Polyelectrolytes onto Carboxymethylated Cellulose Studied by QCM-D. In contrast to SPAR, the QCM-D method was found to be highly sensitive to the amount of water in the cellulose film since it responded strongly to the initial swelling of the film in water. The total frequency shift between the dry state and the wet, fully swollen state of the cellulose film was -760 Hz for the model surface shown in Figure 5. This shift could be separated into two contributions, one major, instant shift caused by viscous losses when the oscillating crystal contacted the fluid and a second, smaller shift related to the slower uptake of water in the cellulose film. The contribution from the bulk fluid could be calculated according to the theory of Kanazawa and Gordon37 and it was thus also possible to calculate the frequency shift related to the swelling

140

Biomacromolecules, Vol. 10, No. 1, 2009

Enarsson and Wågberg

Table 1. Calculated Adsorbed Amounts of CPAM1 and CPAM4 on Carboxymethylated Cellulose with DS ) 0.015

polyelectrolyte

NaCl concentration (M)

adsorbed amount according to SPAR (mg/m2)

CPAM1 CPAM1 CPAM1 CPAM1 CPAM4 CPAM4 CPAM4 CPAM4

10-4 10-3 10-2 10-1 10-4 10-3 10-2 10-1

6.7 7.2 5.3 0.15 1.6 1.4 2.4 1.2

a

QCM mass change due to adsorption (mg/m2)

net change in water content during adsorption, mQCM-mSPAR (mg/m2)

assumed water content in polymer layer (mg/m2) a

estimated total water change in the cellulose film (mg/m2)

6.2 5.5 5.3 0.11 -1.7 -0.11 1.24 0.7

-0.5 -1.7 0.0 0.04 -3.3 -1.5 -1.1 -0.5

31 33 24 1.0 5.1 4.3 7.4 3.8

-31 -34 -24 -1.1 -8.4 -5.8 -8.6 -4.3

Using data on the fraction of water in the polymer layer determined on silica surfaces.29

process. The frequency shift due to viscous losses to the bulk liquid, ∆fbulk, is given by

(

∆fbulk ) -f03 ⁄ 2

ηF πµQFQ

)

1⁄2

(4)

where f0 is the fundamental resonance frequency, η and F are the viscosity and density of the bulk fluid, and FQ and µQ are the density and shear modulus of quartz. Insertion of the values for water at 298 K (η ) 9.12 × 10-4 kg/ms, F ) 997 kg/m3) for a resonator oscillating at 4.95 MHz gave a calculated frequency shift of -670 Hz relative to vacuum. Subtracting this value from the total frequency shift of -760 Hz for the model surface swollen in water gives a net frequency shift of -90 Hz that can be attributed to the uptake of water by the cellulose film. This shift corresponded to a Sauerbrey mass of 16 mg/m2 if viscous losses in the oscillation movement of the cellulose film are neglected. This approximation was motivated by comparing the absolute dissipation of the fully swollen film, 170 × 10-6 units, with that of a bare crystal, which typically is in the range of (168 ( 3) × 10-6 units, which in turn suggests that the model surface could be approximated as an elastic layer. When the initial thickness of 30 nm is considered, this corresponds to a volume swelling of 53% which is close to earlier reported values.21 Apart from the initial swelling of the cellulose film in water, there was clearly a change in the water content of the charged cellulose film when polyelectrolytes were adsorbed, as indicated by the occasional increase in frequency for polyelectrolyte adsorption at low salt concentration. The present data indicate that the net frequency response of QCM was determined by a balance between water released from the cellulose film and the mass increase due to adsorption of polyelectrolytes and immobilized water at the cellulose surface. With respect to swelling, cellulose can theoretically be treated as a polyelectrolyte gel with charges immobilized in a three-dimensional network.38 In an aqueous environment the cellulose charges give rise to an osmotic pressure that leads to a substantial uptake of water, which is strongest under salt-free conditions. It is here suggested that the increase in frequency that was seen at low salt concentration for the adsorption of the highly charged polyelectrolytes CPAM4 and pDADMAC was related to a release of water from the cellulose when the charges were neutralized by an adsorbing polyelectrolyte. Strong deswelling effects have previously been illustrated by Kabanov et al.39 for polyelectrolyte adsorption to a polyacrylate gel. Grimshaw et al.40 have in another work shown that a polyelectrolyte gel can deswell by applying a voltage potential over the network. For pDADMAC adsorption on cellulosic fibers, Swerin et al.41 have shown that the fibers are deswelled in deionized water if pDADMAC is added to the suspension.

To get a decrease in the sensed mass with QCM-D, there were probably two requisites to be fulfilled; first, a large release of water from the cellulose film, and second, a flat conformation of the adsorbed polyelectrolyte layer that would minimize the amount of water entrapped in the polyelectrolyte layer. When salt was added, the degree of swelling of the cellulose film was reduced while the inclusion of water in the adsorbed polyelectrolyte layers was increased, due to an increasing proportion of polyelectrolyte segments in loop and tail configurations.34 These two effects could qualitatively explain why the QCM-D data indicated a mass decrease at low salt concentrations but a mass increase at higher salt concentrations. Structural Model for the Adsorption of Polyelectrolytes on Swollen Cellulose Films. The interpretation of the polyelectrolyte adsorption on cellulose was further developed by directly comparing the adsorbed amounts using the SPAR and QCM models (eqs 1 and 2). Table 1 shows the adsorbed amounts of CPAM1 and CPAM4 obtained with either technique on cellulose with DS ) 0.015. Despite the fact that QCM senses both polyelectrolyte and bound water, the obtained change in Sauerbrey mass was typically smaller than the dry adsorbed amount of polyelectrolyte obtained with SPAR. This suggested that water was generally released from the cellulose film upon polyelectrolyte adsorption, even at salt concentrations up to 10 mM salt. To better understand the deswelling of cellulose, it is of interest to separate the water content into two terms: water in the polyelectrolyte layer and water in the cellulose film. This discrimination can not be done directly based on QCM modeling, but it could be attempted based on earlier SPAR and QCM data. From previous experiments using silica surfaces, it is known that both CPAM1 and CPAM4 form adsorbed layers with a water content.35 Under the assumption that the fraction of water in the adsorbed layer is the same for cellulose and silica substrates, the water content in the polyelectrolyte layer as well as the cellulose film could be calculated. This assumption should be valid for a rigid cellulose film having a charge density similar to silica, provided that the polymer-surface interactions are dominated by electrostatics. Current data supports that electrostatics dominate the adsorption, while the charge density of the cellulose surface (approximately 0.4 µeq/m2) was slightly higher than the reported figure for silica (0.1 µeq/m2 at 1 mM NaCl and pH 7).42 The results are shown in the last two columns of Table 1. As the polyelectrolyte layer was predicted to bring a substantial amount of water to the surface interface, a considerable amount of water was on the other hand predicted to be released from the cellulose film. For CPAM1, the calculations estimated a release of the total water content in the cellulose film, while for CPAM4, the released water was estimated to about 25% of the water in the cellulose film. The

Polyelectrolyte Adsorption on Thin Cellulose Films

reason why CPAM1 with its lower charge density appeared to release more water from the cellulose film compared to CPAM4 is not obvious. It might however be related to the greater mass of CPAM1 that was required to neutralize all the cellulose charges. If one assumes that the polyelectrolyte replaces bound water at the cellulose surface, CPAM1 would replace more water than CPAM4 due to its higher surface coverage.

Conclusions The SPAR and QCM-D methods were complementary because the combination of these enabled the contributions from polyelectrolytes and water in the adsorption to be separated. The advantage of SPAR was that the adsorption signal from the polyelectrolyte was independent of the dynamic swelling processes on the cellulose surfaces, which made it simple to study the influence of polyelectrolyte charge density and salt concentration on the polyelectrolyte adsorption. From the SPAR data it was found that the polyelectrolyte adsorption on the cellulose films was driven mainly by electrostatic interactions since the adsorption was totally eliminated by electrostatic screening at 300 mM NaCl concentration. The saturation point in the adsorption appeared to be closely related to a charge neutralization mechanism in 0.1 mM salt, because all three polyelectrolytes showed a similar number of adsorbed charges. In contrast to SPAR, the QCM-D technique also sensed the water content of the cellulose-polyelectrolyte interface, which was important for studying the state of swelling in the cellulose layer before and after addition of polyelectrolytes. The mass sensed from QCM was concluded to be the difference between water released from the cellulose film and the adsorbed mass of polyelectrolytes including water entrapped in the adsorbed layer. Because the adsorbed polymer mass determined with SPAR was typically greater than the increase in Sauerbrey mass determined with QCM, it was concluded that the cellulose film typically deswelled upon adsorption of polyelectrolytes. Valuable information about the conformation of the adsorbed polyelectrolytes was also obtained from the dissipation shifts. The dissipation appeared to be relatively insensitive to the swelling of the cellulose layer but changed mainly during polyelectrolyte adsorption. The thickest polyelectrolyte layers were obtained at a salt concentration of about 10 mM NaCl, and the adsorption was significantly reduced by electrostatic screening at higher salt concentrations. Acknowledgment. L.E. acknowledges the Swedish Research Council (VR) for funding this study.

References and Notes (1) Kra¨ssig, H.; Schurz, J.; Steadman, R.; Schliefer, K.; Albrecht, W.; Mohring, M.; Schlosser, H. Cellulose. In Ullmann’s Encyclopedia of Industrial Chemistry, 6th ed.; Wiley-VCH: New York, 2004. ¨ dberg, L. AdV. Mater. 1996, (2) Buchholz, V.; Wegner, G.; Stemme, S.; O 8, 399.

Biomacromolecules, Vol. 10, No. 1, 2009

141

¨ sterberg, M. Chem. Soc. ReV. 2006, 35, (3) Kontturi, E.; Tammelin, T.; O 1287. (4) Schaub, M.; Wenz, G.; Wegner, G.; Stein, A.; Klemm, D. AdV. Mater. 1993, 5, 919. ¨ dberg, L.; Rasmusson, J.; (5) Holmberg, M.; Berg, J.; Stemme, S.; O Claesson, P. J. Colloid Interface Sci. 1997, 186, 369. (6) Tammelin, T.; Saarinen, T.; Osterberg, M.; Laine, J. Cellulose 2006, 13, 519. (7) Gunnars, S.; Wågberg, L.; Cohen Stuart, M. A. Cellulose 2002, 9, 239. (8) Geffroy, C.; Labeau, M. P.; Wong, K.; Cabane, B.; Stuart, M. A. C. Colloid Surf., A 2000, 172, 47. (9) Notley, S. M.; Eriksson, M.; Wågberg, L.; Beck, S.; Gray, D. G. Langmuir 2006, 22, 3154. (10) Eriksson, J.; Malmsten, M.; Tiberg, F.; Callisen, T. H.; Damhus, T.; Johansen, K. S. J. Colloid Interface Sci. 2005, 284, 99. (11) Edgar, C. D.; Gray, D. G. Cellulose 2003, 10, 299. (12) Wågberg, L.; Decher, G.; Norgren, M.; Lindstro¨m, T.; Ankerfors, M.; Axna¨s, K. Langmuir 2008, 24, 784. (13) Aulin, C.; Varga, I.; Claesson, P. M.; Wågberg, L.; Lindstro¨m, T. Langmuir 2008, 24, 2509. (14) Decher, G. Science 1997, 277, 1232. (15) Rojas, O. J.; Ernstsson, M.; Neuman, R. D.; Claesson, P. M. J. Phys. Chem. B 2000, 104, 10032. (16) Swerin, A.; Wågberg, L. Nord. Pulp Pap. Res. J. 1994, 9, 18. (17) Terayama, H. J. Polym. Sci. 1952, 8, 243. (18) Horn, D. Prog. Colloid Polym. Sci. 1978, 65, 251. (19) Walecka, J. A. Tappi J. 1956, 39, 458. (20) Katz, S.; Beatson, R. P.; Scallan, A. M. SVen. Papperstidn. 1984, 87, R48. (21) Fa¨lt, S.; Wågberg, L.; Vesterlind, E. L.; Larsson, P. T. Cellulose 2004, 11, 151. (22) Dijt, J. C.; Stuart, M. A. C.; Fleer, G. J. AdV. Colloid Interface Sci. 1994, 50, 79. (23) Torn, L. H.; Koopal, L. K.; de Keizer, A.; Lyklema, J. Langmuir 2005, 21, 7768. (24) Sennerfors, T.; Bogdanovic, G.; Tiberg, F. Langmuir 2002, 18, 6410. (25) Lingstro¨m, R.; Wågberg, L.; Larsson, P. T. J. Colloid Interface Sci. 2005, 296, 396. (26) Sauerbrey, G. Z. Phys. 1959, 155, 206. (27) Edvardsson, M.; Rodahl, M.; Kasemo, B.; Ho¨o¨k, F. Anal. Chem. 2005, 77, 4918. (28) Rodahl, M.; Kasemo, B. Sens. Actuators, A 1996, 54, 448. (29) Enarsson, L.-E.; Wågberg, L. J. Colloid Interface Sci. 2008, 325, 84. (30) Fleer, G. J.; Cohen Stuart, M. A.; Scheutjens, J. M. H. M.; Cosgrove, T.; Vincent, B. Polymers at Interfaces, 1st ed.; Chapman and Hall: New York, 1993. (31) Wågberg, L. Nord. Pulp Pap. Res. J. 2000, 15, 586. (32) Horvath, A. E.; Lindstro¨m, T.; Laine, J. Langmuir 2006, 22, 824. (33) Pelton, R. H. J. Colloid Interface Sci. 1986, 111, 475. (34) Shubin, V.; Linse, P. J. Phys. Chem. 1995, 99, 1285. (35) Enarsson, L.-E.; Wågberg, L. Langmuir 2008, 24, 7329. (36) Trout, P. E. Tappi 1951, 34, 539. (37) Kanazawa, K. K.; Gordon, J. G. Anal. Chem. 1985, 57, 1770. (38) Grignon, J.; Scallan, A. M. J. Appl. Polym. Sci. 1980, 25, 2829. (39) Kabanov, V. A.; Zezin, A. B.; Rogacheva, V. B.; Prevish, V. A. Makromol. Chem. 1989, 190, 2211. (40) Grimshaw, P. E.; Nussbaum, J. H.; Grodzinsky, A. J.; Yarmush, M. L. J. Chem. Phys. 1990, 93, 4462. ¨ dberg, L.; Lindstro¨m, T. Nord. Pulp Pap. Res. J. 1990, (41) Swerin, A.; O 5, 188. (42) Bolt, G. H. J. Phys. Chem. 1957, 61, 1166.

BM8010193