Electrophoresis 2009, 30, 18771887

1877

Malinda Salim1 Phillip C. Wright1 Sally L. McArthur2

1

Research Article

ChELSI, Department of Chemical and Process Engineering, The University of Shefeld, Shefeld, UK 2 IRIS, Faculty of Engineering and Industrial Sciences, Swinburne University of Technology, Hawthorn, Victoria, Australia

Studies of electroosmotic ow and the effects of protein adsorption in plasmapolymerized microchannel surfaces
This paper presents a study of EOF properties of plasma-polymerized microchannel surfaces and the effects of protein (brinogen and lysozyme) adsorption on the EOF behavior of the surface-modied microchannels. Three plasma polymer surfaces, i.e. tetraglyme, acrylic acid and allylamine, are tested. Results indicate EOF suppression in all plasma-coated channels compared with the uncoated glass microchannel surfaces. The EOF behaviors of the modied microchannels after exposure to protein solutions are also investigated and show that even low levels of protein adsorption can signicantly inuence EOF behavior, and in some cases, result in the reversal of ow. The results also highlight that EOF measurement can be used as a method for detecting the presence of proteins within microchannels at low surface coverage (o1 ng/cm2 on glass). Critically, the results illustrate that the non-fouling tetraglyme plasma polymer is able to sustain EOF. Comparison of the plasma-polymerized surfaces with conventionally grafted polyelectrolyte surfaces demonstrates the stabilities of the plasma polymer lms, enabling multiple EOF runs over 3 days without deterioration in performance. The results of this study clearly demonstrate that plasma polymers enable the surface chemistry of microuidic devices to be tailored for specic applications. Critically, the deposition of the non-fouling tetraglyme coating enables stable EOF to be induced in the presence of protein. Keywords: EOF / Microchannel / Plasma polymerization / Protein adsorption DOI 10.1002/elps.200800619

Received June 17, 2008 Revised December 28, 2008 Accepted January 5, 2009

1 Introduction
There has been a growing interest in the development of CE and other analytical techniques capable of separating proteins and peptides in free solution under non-denaturing conditions [1, 2]. CE has typically been performed in glass tubes or capillaries, and more recently transferred to microuidic chips for rapid separation [3, 4]. One of the
Correspondence: Dr. Sally L. McArthur, IRIS, Faculty of Engineering and Industrial Sciences, Swinburne University of Technology, Mail H66, 543-545 Burwood Road, P. O. Box 218, Hawthorn, Victoria 3122, Australia E-mail: smcarthur@swin.edu.au Fax: 161-3-9214-5050

Abbreviations: EDC, 1-ethyl-3-(3-dimethylaminopropyl) electroosmotic carbodiimide hydrochloride; lEOF, mobilities; NHS, N-hydroxysuccinimide; PAH, polyallylamine hydrochloride; PEI, polyethyleneimine; pp.AAc, plasma-polymerized acrylic acid; pp.AAm, plasmapolymerized allylamine; pp.TG, plasma-polymerized tetraglyme; XPS, X-ray photoelectron spectroscopy

challenges in CE, regardless of the platform, is the interaction of proteins and peptides in the solution with the wall surfaces [5, 6]. Therefore, attempts to reduce the extent of non-specic proteinsurface interaction have been widely reported [7, 8]. However, conditions where the proteinsurface interaction is minimized do not always result in optimal conditions for the separation system. Other factors, including EOF that arises from the surface charge, also need to be considered [9, 10]. EOF is the bulk ow of uid through the channel brought about by an applied electric eld. It can be used for the separation of charged species [9, 10], and also as a method of surface characterization to determine the surface isoelectric point and monitoring surface coating stability [11]. Variations in EOF are prevalent in microuidic devices manufactured from inorganic substrates such as glass, quartz and fused silica. Thus, there is a need to understand the factors that can induce variations and inconsistencies in EOF that may then interfere with device performance. In common with protein adsorption, EOF is dependent on solution conditions and surface chemistries. Furthermore, the adsorption of proteins on the surfaces may
www.electrophoresis-journal.com

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1878

M. Salim et al.

Electrophoresis 2009, 30, 18771887

change the surface charge and affect the EOF behavior, adding further complexity to the system. Generally, adsorption of proteins on surfaces is controlled by changing the system solution conditions (pH, ionic strength) and/or modifying the surface chemistry using polymers or surfactants [12, 13]. The effects of solution conditions on the adsorption behavior of proteins in micro-dimensional channels have been reported using ELISA [14], uorescence [15] and radiolabeling [16] methods. Two widely used strategies for modifying microchannel surfaces are dynamic or physical adsorption and covalent grafting of polyelectrolytes [7, 17, 18]. Both techniques have their unique advantages and disadvantages. Physical adsorption, for example, despite its easier application, often suffers from a lack of stability and may alter system selectivity. Although covalently grafting improves the stability, it often requires complicated procedures and can be time- consuming. Difculties in surface modication also often arise due to the large variations in chip construction materials [7]. Since commonly used solvent or liquid-phase coatings may cause problems, such as clogging or poor uniformities in the microchannels, solventless deposition (plasma and non-plasma chemical vapor deposition) has also been sought as an alternative method to coat microchannels [19]. As an example, Kadowaki et al. and Yoshiki et al. coated the inner surface of microcapillaries using a microplasma [20, 21]. In 2003, the Karube group was the rst to investigate the use of plasma polymerization to coat open glass microchannels, comparing the effects of hydrophilic and hydrophobic coatings for IEF-based protein separation [22, 23]. Barbier et al. introduced plasma-polymerized acrylic acid (pp.AAc) on PDMS microchannels and showed stability over several days [24]. These early studies highlight that plasma polymerization may be a versatile method for coating microchannel surfaces, where it can be used to deposit thin lms of various structures and functionalities on almost any substrates without altering the bulk properties. Previously, we have successfully demonstrated plasma polymerization of low fouling tetraethylene glycol dimethyl ether (plasma polymerised tetraglyme, pp.TG) onto glass microuidic channels and PTFE surfaces [25]. This paper aims at extending this work to investigate the electroosmotic behavior of several commonly used plasma polymers: pp.TG, plasma-polymerized allylamine (pp.AAm) and pp.AAc. pp.AAm and pp.AAc surfaces are commonly utilized for the immobilization of biomolecules and grafting of polymers [2629] and produce surfaces rich in amines and acids, respectively. These systems allow us to compare the stability and EOF behavior of the plasma-polymerized coatings with the coatings produced conventionally using physical adsorption and covalent grafting of polyelectrolytes. Once the coating stability was established, it was possible to explore the inuence of protein adsorption on surfaces of differing chemical characteristics. The pp.TG coating was used as a low fouling surface to establish if it is possible to create stable and reproducible EOF in the presence of proteins.
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

2 Materials and methods

2.1 Chemicals and materials Fibrinogen from human plasma (65% purity, contains 15% sodium citrate and 20% sodium chloride), polyethyleneimine (PEI) solution (MW: 750 000 g/mol, 50% w/v in H2O), polyallylamine hydrochloride (PAH, MW: 15 000 g/mol), allylamine (98% purity), acrylic acid (99% purity), 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), N-hydroxysuccinimide (NHS), lysozyme from chicken egg white (MW: 14 700 g/mol, Z90% purity), c-phycocyanin from Spirulina sp., myoglobin from equine heart, and DMSO were all purchased from Sigma-Aldrich (UK). Distilled water was deionized (ddH2O, 18 mO) using a Milli-Q system from Millipore (UK). One milliliter disposable syringes and pH meter (Fisherbrand Hydrus 300) were purchased from Fisher Scientic (UK). One hundred microliter Hamilton microsyringes were obtained from Thames Restek (UK). Citrate buffer (pH 3.0 and 4.4), (PBS, pH 5.9, 7.4 and 8.3) and carbonate buffer (pH 9.2 and 9.9) of constant ionic strengths were used in the current-monitoring measurements; 20 mM NaOH and 10 mM H3PO4 solutions were all prepared using ddH2O. Solutions were degassed by sonication and lter-sterilized using 0.2 mm cellulose acetate membranes (Presearch, UK). PTFE sheets were obtained from the University of Shefeld main workshop and cleaned by sonicating in ddH2O for 15 min prior to use. All the chemicals used for buffer preparation were purchased from Sigma-Aldrich and used without further purication.

2.2 Microuidic chip A linear microuidic channel (2.2 cm long) was prepared in glass using a photo-chemical etching process. This involved the transfer of a pattern onto a photo-resist mask deposited on a glass substrate and subsequent isotropic etching using hydrouoric acid to the required depth. A D-shaped etched channel was formed. Each end of the microchannel was drilled to form a reservoir (3 mm od), and the glass chip was then placed with the etched channel facing down on top of a PTFE gasket. The composite device was then clamped within a chip holder to form a seal. The width (measured on top of the channel) and depth of the microchannel were 150 and 75 mm, respectively, as measured using an optical microscope (Axiovert 100, Carl Zeiss, USA).

2.3 Plasma polymerization Plasma polymerization is the formation of a polymeric lm from the vapor phase under the inuence of plasma [30]. Volatilized monomers are introduced into an evacuated reaction chamber where, under the inuence of an electric
www.electrophoresis-journal.com

Electrophoresis 2009, 30, 18771887

Microuidics and Miniaturization 2.5 XPS

1879

eld, they undergo ionization, generating electrons, ions, free radicals, photons and molecules both in ground and in excited states. Excitation of the monomer results in reactive species on the surface, which act as reactive sites for the covalent attachment of other species [30]. The schematic diagram of a typical radio-frequency glow discharge plasma reactor is depicted in Supporting Information Fig. A. Although the specics of the reactor designs varied slightly with each monomer, the basic components of the allylamine and acrylic acid systems remained constant, while the tetraglyme system was signicantly different. Details of the systems can be found in our earlier papers [25, 31]. In all cases both the glass microuidic channels and the PTFE gaskets were simultaneously coated with the plasma polymer. The deposition conditions used for each monomer are summarized in Table 1. All these deposition times and conditions were chosen so that the plasma polymers retained a high level of functionality, were stable to rinsing and had sufcient thickness to screen the underlying substrate signal during X-ray photoelectron spectroscopy (XPS) (i.e. thicker than 10 nm).

2.4 Physical adsorption and covalent grafting of PEI and PAH on pp.AAc surfaces Solutions containing 0.3% PEI and PAH (3 mg/mL in 150 mM PBS, pH 7.4) were prepared and pH adjusted to 7.0 and 5.0, respectively, with 1.0 M HCl. These pH values were chosen because the polymers are highly dissociated under these conditions [32, 33]. The pp.AAc-coated microchannels and PTFE sheets were soaked in PEI and PAH solutions for 30 min at room temperature, then rinsed with PBS and a large volume of ddH2O to remove loosely bound polymers. To covalently immobilize the polyelectrolytes, stock solutions (10 ) of 75 mM EDC and 15 mM NHS crosslinkers were prepared, i.e. 0.114 g EDC, 10 mL NHS (0.0172 g NHS in 100 mL DMSO) and 1 mL PBS (150 mM, pH 7.4). The pp.AAc-coated microchannels and PTFE sheets were activated by adding solutions of the EDC and NHS (1 ), followed by incubation at room temperature for 40 min [34]. Samples were then immediately transferred and submerged in 0.3% PEI and PAH solutions and incubated overnight at room temperature. After incubation, samples were rinsed thoroughly with PBS and ddH2O, and EOF mobility measurements performed.

XPS analysis was performed using a Kratos AXIS Ultra DLD instrument equipped with a monochromatic Al Ka X-ray source operated at a power of 150 W, with charge compensation on. A take-off angle of 01 relative to the sample normal was used for all measurements. Survey spectra (12000 eV) were collected at 160 eV pass energy with a 1 eV step width. The pressure in the main ultra-high vacuum chamber was maintained below 1 108 mbar for all analyses. Data analysis and charge correction were carried out using CasaXPS software Version 2.2.107, with the C 1s component of the aliphatic hydrocarbon set to 285.0 eV as an energy reference and the relative sensitivity factors supplied with the instrument. High-resolution C 1s core level spectra were collected at 20 eV pass energy and a 0.1 eV step width. Linear background was applied to the spectrum, and the component curves tted to estimate the amount of different carbon functionalities present on the surface. The peaks tted were GaussianLorentzian (30GL). The carbon functionalities have different binding energies with certain chemical shifts, and the C 1s chemical shifts relative to hydrocarbon were obtained from the ESCA300 database [35]. The full-widthhalf-maximum for unresolved peaks was constrained to be equal to that of the hydrocarbon peak.

2.6 EOF measurement A current-monitoring method was used to determine the electroosmotic mobilities (mEOF) in the microchannels [36]. A high-voltage supply (Paragon 3B PSU, Kingeld Electronics, UK), having adjustable voltage from 0 to 4000 V DC, was used to apply the electric eld across the microchannel, which was connected to platinum wire electrodes. Changes in current were monitored by measuring the voltage drop across a 10 kO resistor, with the signal temporally recorded with an in-house program written in Lab-View (Version 8, National Instruments, UK). The mEOF measurements as a function of pH were carried out using 20 and 18 mM PBS, citrate and carbonate buffers as the higher and lower ionic strengths, respectively (10% ionic strength difference). A buffer ionic strength of 20 mM was selected for this study, as it is the common concentration used in CE applications [37]. Prior to mEOF measurements, the microuidic channels were rinsed with 1 mL of 0.1 M NaOH solution and

Table 1. Deposition conditions used for plasma polymerization Monomer Structure Deposition power (W) Tetraglyme Acrylic acid Allylamine CH3[OCH2CH2]4OCH3 CH2= =CHCOOH CH2= =CHCH2NH2 10 10 10 Deposition conditions Deposition time (min) 25 15 15 Monomer owrate (sccm) 2.5 2.2 2.2

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.electrophoresis-journal.com

1880

M. Salim et al.

Electrophoresis 2009, 30, 18771887

incubated for 10 min, followed by 3 mL of ddH2O rinsing. This pre-conditioning of the non-coated glass surfaces is usually required to ensure that most of the surface silanol groups are dissociated [38]. Rinsing was performed by pressure pumping through a 1 mL disposable syringe. The specied buffer of lower conductivity, i.e. 18 mM, was then owed through the chip and incubated for 2 min at room temperature (10 min for uncoated chips). A longer buffer incubation time, i.e. 10 min, was found to be necessary in order to obtain reproducible EOF runs in uncoated chips. This can be due to the slow equilibration of the surface charge in the glass surface [39]. After buffer incubation, the lower conductivity buffer solution was removed from one of the reservoirs and replaced by slowly lling the reservoir with a higher conductivity buffer of equal volume to minimize errors induced by hydrostatic pressure. Changes in current were observed upon voltage application due to the buffer replacement that arises from EOF. The time required for the current to reach plateau was recorded, and the mEOF calculated as follows: L 1 mEOF tE where mEOF is the electroosmotic mobility (cm2/V s), L the length of the microchannel (2.2 cm), E the applied electric eld strength (68 V/cm) and t the time for buffer replacement or, in other words, the time taken for current to reach a plateau (in seconds). The applied electric eld strength of 68 V/cm was selected for use in this study to minimize the inuence of Joule heating on the measured electroosmotic mobility values, i.e. the selected value falls in the region where little deviation from linearity was observed in the electroosmotic velocity versus the applied electric eld strength prole (see Supporting Information Fig. B). For the protein adsorption experiments, human brinogen solution (0500 mg/mL) and chicken egg lysozyme solution (0500 mg/mL) were prepared in 150 mM PBS, pH 7.4, and incubated in the microchannel for 30 min at room temperature. The electroosmotic mobilities (mEOF) were measured using the current-monitoring method described in Section 2.6.

(assuming that of water for low buffer ionic strengths, 0.001 kg/m s). The streaming potential apparatus was based on the design of Van Wagenen, as modied by Scales et al. [41]. In brief, the polymer-coated glass slides were tted into channels within a pair of Teon blocks. The two blocks were then clamped together, and a capillary formed between the two polymer-treated surfaces. An electrolyte of a known pH and conductivity was then pumped through the capillary, and variations in pressure and voltage across the capillary were measured. All measurements were performed between polymer-coated glass slides with 1 mM NaNO3 solutions. In order to ascertain the isoelectric point for each surface, the pH of the electrolyte was varied between 2 and 11 using 0.1 M HNO3 and 0.1 M NaOH. The streaming potential was measured for a minimum of three cycles at each pH and then converted to an average zeta potential using the following Smoluchowski equation [42]: DP e0 er z 3 Streaming potential Zk where DP is the pressure drop across the system (N/m2), z the zeta potential (V), e0 the electrical permittivity of a vacuum (F/m), er the relative dielectric constant, Z the solution viscosity (kg/m s) and k the specic conductance of the solution ( m).

3 Results and discussion

3.1 XPS Prior to the EOF measurements, the surface chemical compositions of the plasma polymers were characterized using XPS (see Table 2). Analysis of the pp.AAc-coated glass surface showed the presence of oxygen (2673%) and carbon (7578%), with no substrate silicon signal observed, indicating that the coating was thicker than 10 nm. Five component peaks were tted to the C 1s prole, which comprised 5175% hydrocarbon (CR) at 285.0 eV, 1772% hydroxyl/ether (CO) at 286.6 eV, 1271% carboxyl/ester (COO) at 289.4 eV with corresponding b shift (CCOO) at 285.7 eV and 871% carbonyl (C= =O) groups at 287.9 eV. The elemental composition of the pp.TG-coated microchannel surface obtained from the wide scan analysis indicated the presence of a coating with 3173% oxygen and

2.7 Zeta potential calculations and streaming potential measurements Assuming a thin electrical double layer (a typical 1 nm1 mm) with large channel height, zeta potential calculations can be determined using the HelmholtzSmoluchowski equation as shown in the following equation [40]: m Z z EOF 2 e0 er where z is the zeta potential (V), e0 the electrical permittivity of a vacuum (8.85 1012 F/m), er the relative dielectric constant (78.4) and Z the solution viscosity
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Table 2. Percentage atomic concentrations from XPS wide scan

analysis of pp.AAc, pp.TG and pp.AAm surfaces

Surfaces O 1s pp.AAc pp.TG pp.AAm 2673 3173 571

Atomic percentage of elements (%) C 1s 7578 6977 8278 N 1s 1371

www.electrophoresis-journal.com

Electrophoresis 2009, 30, 18771887

Microuidics and Miniaturization

1881

6977% carbon (Table 2). This is as expected from the tetraglyme monomer (CH3[OCH2CH2]4OCH3), as it comprises only oxygen and carbon apart from hydrogen, which is not detectable by XPS. Three components were tted into the C 1s spectra of pp.TG surfaces (Fig. 1B), with hydroxyl/ether (C= =O) forming the major constituent (8478%) at 286.6 eV relative to hydrocarbon (CR) at 285.0 eV (1471%). Carbonyl groups (C= =O) were also found to be present at 289.2 eV at 270%. The XPS wide scan analysis of the pp.AAm surface revealed the presence of carbon, nitrogen and oxygen (8278, 1371 and 571%, respectively). Although the starting monomer (CH2= =CHCH2NH2) does not contain oxygen, this is likely due to post-plasma oxidation that is known to be associated with amine containing plasma polymers [43]. Analysis of the high-resolution XPS C 1s spectra of pp.AAm-coated surfaces (Fig. 1C) indicated four components present in the spectra, with 4875% of hydrocarbon (CH), 1271% amines (CN), 2873% hydroxyl or

ether/imide groups (CO/C= =N) and 1171% carbonyl/ amide groups (C= =O/NC= =O) observed. The nitrogen and oxygen groups present in pp.AAm make curve tting difcult, as CO and CN peaks overlap in the C 1s prole. The complex nature of the plasma polymerization process means that the surface may contain many nitrogencontaining species, with imines, amides and nitrile groups in addition to the desired amines [31].

3.2 EOF measurements 3.2.1 Plasma-polymerized surfaces The EOF of uncoated and plasma-polymerized channels was measured as a function of buffer pH from 3.0 to 9.9. Results are shown in Fig. 2. The magnitude of mEOF was smaller in all plasma-polymerized surfaces compared with the uncoated glass microchannels. Coating of the microchannels resulted in both EOF suppression and more reproducible mEOF runs. Uncoated pp.TG and pp.AAc-coated microchannels all displayed cathodic ow over a large pH range, with the highest mEOF observed in uncoated channels at pH 7.4 (4.9 104 cm2/V s), decreasing to 1.3 104 and 1.7 104 cm2/V s for pp.TG and pp.AAc surfaces, respectively. A shift in the surface isoelectric point was also observed for pp.TG and pp.AAc-coated surfaces from pH o3.0 (non-coated channel) to pH 3.04.4. The negatively charged pp.TG surface was a little surprising, but can be attributed to a combination of factors. In the XPS C 1s data, there is some evidence of a binding energy shift to energies indicative of carboxyl species. In addition, a derivatization study (data not shown) has suggested that a small proportion of the CO species attributed to the ether carbons (286.5 eV) are in fact hydroxyl groups generated during the plasma process. Previous studies on ethylene oxide terminated self-assembled monolayers have suggested that there is some level of hydroxyl group deprotonation in an aqueous environment [44]. The mEOF was almost constant from pH 5.9 to 9.9,

A
CH

C=O/O-C-O C*-COO C-C*OOR

B
C-O

C-H C=O

C
CH

C-N C-O/C=N C=O/CNO

Figure 1. XPS C 1s spectra of plasma-polymerized acrylic acid (A), tetraglyme (B) and allylamine (C) coated surfaces.

Figure 2. Electroosmotic mobilities of non-coated microchannel, pp.TG, pp.AAc and pp.AAm-coated channel as a function of buffer pH 310. Each point represents an average of three independent EOF measurements.

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.electrophoresis-journal.com

1882

M. Salim et al.

Electrophoresis 2009, 30, 18771887

indicating a complete dissociation of the relatively small number of hydroxyl and acidic groups. The mEOF of the pp.AAc surface exhibited cathodic ow from pH 4.4 to 9.9, indicating the presence of excess negative charges on the surface. The negative surface charge is the result of the dissociation of the acidic groups present, which increases with pH values, with highest degree of protonation at pH 9.9 with maximum charge density. The results illustrated in Fig. 2 also show similarities between the EOF behavior of pp.AAc and pp.TG surfaces. Deposition of pp.AAm on microchannel surfaces shows a completely different mEOF behavior. A reversed EOF direction (anodic ow) was observed from pH 3.0 to 8.3, with the mEOF ranging from 1.5 104 cm2/V s at pH 3.0 to 3.8 105 cm2/V s at pH 9.9. This is due to the protonation of amine groups at low pH, which results in a positively charged surface. Above that pH, the surface was found to be slightly negatively charged. The absence of mEOF was observed at pH 8.39.2, which indicates the surface isoelectric point. While the allylamine is a primary amine, having the monomer structure of CH2= =CHCH2NH2, the resulting plasma polymer does not reproduce the monomer chemistry. Differentiation between primary, secondary and tertiary forms of the amine groups on the pp.AAm surface, which have close pKa values [45], is not possible from the data obtained in this study, as the nitrogen is combined in the lms as imines and amides as well as the variety of amine forms [31]. All these plasma-polymerized surfaces were shown to be relatively stable over the pH range tested (pH 3.09.9), which enabled EOF runs with good reproducibility over a minimum of 3 days (EOF experiments were repeated each day to conrm reproducibility, which was within 75%). However, the pp.AAm surface was observed to require a longer time for the EOF to stabilize from a freshly prepared coating, i.e. about 35 EOF runs compared with the rst 12 runs for pp.TG and pp.AAc. This observation may be due to the removal of soluble low-molecular-weight fragments from the surface and/or post-plasma oxidation in pp.AAm [46]. Plasma-polymerized surfaces were also stable to rinsing with 0.1 M NaOH, although surface pre-treatment was not necessary to obtain reproducible EOF. 3.2.2 Wet chemistry PAH and PEI are weak cationic polyelectrolytes used largely in layer-by-layer deposition of charged coatings [47]. PEI is a highly branched polymer, which consists of primary, secondary and tertiary amines with a ratio of 1:2:1 (pKa value 10.011.0) [48]. PAH, however, is a linear polymer having primary amines on the chain (pKa 8.5) [33]. A comparison was made between the EOF behavior of pp.AAm surfaces and surfaces made by the physical adsorption and covalent grafting of PAH and PEI. In the physical adsorption studies, both PAH and PEI solutions were prepared in 150 mM PBS and adsorbed on pp.AAc surfaces at pH 5.0 and 7.0, respectively [32, 33]. Surfaces
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

formed in high-molarity buffer were expected to be more homogeneous, as small ions may interact with the adsorbed polymer lms, which result in structural changes of the resulting lms [47]. It should be noted that the structures of these adsorbed polyelectrolyte lms are strongly dependent on the deposition conditions, particularly the pH and ionic strengths of the buffer solutions. Other than the adsorption conditions, the effects of pH of the conditioning solutions are also known to be important on the polyelectrolyte structures and their degree of dissociation [47]. Therefore, changes in mEOF with changes in the pH of the conditioning solutions are to be expected. The effects of pH of the conditioning solutions on mEOF in physically adsorbed PAH and PEI surfaces as well as pp.AAm are shown in Fig. 3. Physical adsorption of PAH and PEI onto the oppositely charged pp.AAc surface is assumed to be predominantly driven by electrostatic interaction. These three coatings exhibited anodic ow over a wide range of pH, indicating that the surfaces were positively charged. The magnitude of mEOF of all these surfaces was smaller than the uncoated microchannel. The positively charged surface of these physically adsorbed polymers arises from the protonation of the amine groups of the PEI and PAH attached to the channel surface. Although the isoelectric points of these three surfaces fell in the pH range of 8.39.2 with similar EOF behaviors, slight differences in the magnitude of mEOF as a function of pH were observed especially at lower pH values. These may be attributed to the differences in the conformation of the polymers at the surface. For instance, the magnitude of mEOF of physically adsorbed primary amines, PAH, was generally smaller at lower pH values, i.e. (7.171.1) 105 cm2/V s compared with (1.370.3) 104 cm2/V s and (1.570.3) 104 cm2/V s for physically adsorbed PEI and pp.AAm, respectively, at pH 3.0. Although allylamine was used as the monomer in plasma polymerization, the EOF properties of pp.AAm surfaces were shown to be different from the physically adsorbed polyallylamine onto the surface. During plasma polymerization, fragmentation of the allylamine monomer results in the primary amines being converted to other

Figure 3. Electroosmotic mobilities of pp.AAm, physically adsorbed PEI and physically adsorbed PAH-coated channel as a function of buffer pH 310. Each point represents an average of three independent EOF measurements.

www.electrophoresis-journal.com

Electrophoresis 2009, 30, 18771887

Microuidics and Miniaturization

1883

complex nitrogen containing functionalities with branching and cross-linking exists randomly within the structure [48]. It can therefore be deduced that surfaces produced by plasma polymerization, which have different structural properties and chemical architecture to the conventional polymers, can inuence the EOF behavior. Despite the simplicity of the physical adsorption methods, it is well known that this type of coating has limited stability with coating detachment only after 420 EOF runs [49]. We observed the coatings of these physically adsorbed polymers deteriorated after the second to third day of runs (30 runs each day). This method is therefore not very suitable for long-term device usage unless the coating is replenished repeatedly, which may be time consuming. It is also not practical for applications requiring mass spectrometry, as analysis may be disrupted and samples contaminated by the polymer released from the microchannel walls. In previous studies, covalent immobilization techniques have been shown to improve the coating stability [49]. The mEOF of covalently grafted PEI and PAH on pp.AAc surfaces by EDC/NHS chemistry are shown in Fig. 4. Results clearly show the presence of PAH and PEI on the pp.AAc surfaces. The covalently grafted PAH surface was negatively charged at pH 9.2 and 9.9 ((2.670.46) 105 and (2.270.0) 105, respectively); while the PEI surface was positively charged over all the pH values tested (surface isoelectric points of PAH and PEI span between the range of pH 8.39.2 and pH49.9, respectively). The higher isoelectric point seen with the covalently coupled PEI surfaces may be due to the generally lower pKa values of primary amines compared with secondary amines, as the alkyl group is slightly electron donating. The intermolecular hydrogen bonding in the amines also results in a decreased electron density of the nitrogen, and lowering its pKa [50, 51]. Another interesting observation from these studies was the difference in mEOF properties of physical adsorption of PEI and covalently coupled PEI at high pH. These observations may arise from the difference in the polymer conformations at the interface. Theoretically, amine groups are deprotonated or lose their charges when pH of the conditioning

solutions approaches the pKa, bearing in mind that PEI consists of different types of amines that have different pKa values. In the physical adsorption process of PEI, primary and secondary amines may be randomly attached to the surface via electrostatic interaction, while primary amines were coupled via the EDC/NHS cross-linking agent to the surface. The resulting conformation of polymers formed by physical adsorption and covalent grafting may thus be different. However, possibilities of polymer detachment from the surface cannot be excluded. Although it is difcult to tell or separate the polymers being physically adsorbed or covalently coupled, the improved stability issues in the covalently grafted polymers (43 days) may actually suggest that there are chemical linkages. Despite higher stability of covalently grafted PAH, the mEOF as a function of pH did not show signicant differences to that of physically adsorbed PAH.

3.3 Surface zeta potential Measurement of a surface potential is difcult and is often estimated based on zeta potential, i.e. potential at the shear plane [52]. On the basis of HelmholtzSmoluchowski equation (see Eq. 2), it is evident that the data obtained from electroosmotic mobility measurements can be used to estimate surface zeta potential. Zeta potential is a useful parameter to characterize surface charge. Theoretically, if vEOF is directly proportional to E, the surface zeta potential is independent of the applied electric eld strength. In practice, however, the vEOF is not a linear function of E due to the increased heat generated at high voltages [53]. Thus, the calculated zeta potential is expected to depend on the applied electric eld strengths to a certain extent [53]. It should also be noted that the solution viscosity and its dielectric constant are temperature dependent [54]. Since estimated values for water solution are used in our calculation, the surface zeta potential calculated from the HelmholtzSmoluchowski relation (Eqs. 2 and 3) is therefore an approximation at best until all the relevant parameters can be addressed. Streaming potential measurements have always been reported as being accurate in determining surface zeta potentials [52]. Unlike EOF, where the solution is moved relative to a charged surface by an applied electric eld, in the streaming potential method, the solution is owed over the surface using applied pressure generating the surface potential [52]. Surface zeta potentials obtained from streaming potential measurements for non-coated glass cover slips over a range of pH at constant ionic strength are shown in Fig. 5. It can be observed that the surface zeta potentials obtained from both methods differ, although the trend is similar. The magnitude of zEO was greater than zSP over the whole pH range tested, with an increasing gap as the pH increased. This observation was in line with those reported in the literature, where zEO4zSP [55, 56]. Differences in the zeta potential values are not unlikely since several potential
www.electrophoresis-journal.com

Figure 4. Electroosmotic mobilities of pp.AAm, EDC/NHSgrafted PEI and EDC/NHS-grafted PAH-coated channels as a function of buffer pH 310. Each point represents an average of three independent EOF measurements.

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1884

M. Salim et al.

Electrophoresis 2009, 30, 18771887

Figure 5. Surface zeta potential of non-coated microchannel as a function of 1 mM buffer pH 310 obtained from electroosmotic mobility measurement (EO) and streaming potential measurement (SP).

errors can arise from both techniques: overestimation of zEO in the presence of nite Joule heating and underestimation of zSP if the surface conductance is ignored [42]. Several authors have also suggested that higher zEO was contributed to the nearer shear plane location to the surface [56, 57]. Thus, as reported by Werner et al., different method comparisons can be used to assess differences in the location of the inner and outer Helmholtz planes as well as the shear plane [58]. However, the exact location of the shear plane is difcult to be determined, particularly in cases where there is a diffuse interface (i.e. grafted polymer lms). This study also conrms that the surface isoelectric point remained similar (opH 3.0) in both methods. Despite these classical techniques, several authors have also proposed new improved methods to obtain more accurate estimates of surface zeta potentials. For example, Erickson et al. suggested an improved slopeintercept technique for streaming potential data [59]. Sze et al. measured the surface z by combining the Smoluchowski equation (Eq. 2) with the measured slope of currenttime relationship in EOF [60]. This technique alleviates the common problem encountered in the current-monitoring method, i.e. difculty in determining the exact time required for buffer replacement.

30-min incubation time was selected since it is common in CE applications and is known to result in a signicant level of protein adsorption (corresponds with the protein adsorption plateau as shown by Salim et al. [14]). The results of these studies are shown in Fig. 6. Positive mEOF represents a cathodic EOF and thus a negatively charged surface. Similarly, a positively charged surface is represented by a negative mEOF. When the net charge of the surface is zero, no mEOF is present. At the experimental condition under investigation here (pH 7.4), human brinogen as well as non-coated, pp.TG and pp.AAc-coated microchannels are net negatively charged. When the concentration of brinogen is increased, the mEOF decreased with a sharp drop between 5 and 50 mg/mL and then reached a plateau region where the mEOF remained almost unchanged with increasing brinogen solution concentrations (Fig. 6). This may indicate that the adsorption of brinogen in uncoated microchannels reached a saturation level at between 50 and 500 mg/mL concentration, which correlates well with data obtained previously on brinogen adsorption in glass microcapillaries [14]. It can be hypothesized that adsorption of brinogen on the glass microchannel surfaces shields the silanol-charged groups, changing the net charge or zeta potential of the surface, thus changing its EOF characteristics [61]. The drop in mEOF for pp.AAc when the brinogen solution concentration increased indicates that brinogen adsorbed on the acid-containing surface, reaching a saturation level at brinogen concentrations between 50 and 500 mg/mL. The amine containing pp.AAm surface was positively charged at pH 7.4 (negative mEOF). Anodic ow decreased when brinogen was introduced to the system. At high brinogen concentrations of 50500 mg/mL, sufcient brinogen was adsorbed on the pp.AAm surface to reverse the ow direction from anodic to cathodic, with the surface becoming negatively charged as a result of protein adsorption. Therefore, when acidand amine-charged surfaces are used for specic or afnity proteins separation, the resulting EOF can be controlled

3.4 Effects of protein adsorption on EOF Since EOF is an important parameter in the application of CE for protein and peptide separation, an understanding of the effects of protein adsorption on ow characteristics is invaluable. Several papers have shown that EOF changes when proteins are present and adsorbed on the surface of a device [40, 61, 62]. These changes, however, vary with the experimental conditions, such as the microchannel material, buffer solutions and protein adsorption conditions. In this study, the effects of human brinogen adsorption on the EOF behaviors of uncoated microchannels, pp.TG, pp.Aam and pp.AAc surfaces were tested. Fibrinogen solution concentrations were varied from 0 to 500 mg/mL and the
& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Figure 6. Electroosmotic mobilities of non-coated, pp.TG, pp.AAc, and pp.AAm-coated microchannels that were incubated with brinogen solutions (0500 mg/mL) for 30 min at pH 7.4. Each point represents an average of three independent EOF measurements.

www.electrophoresis-journal.com

Electrophoresis 2009, 30, 18771887

Microuidics and Miniaturization

1885

by varying the amount of protein introduced, and of course, the solution conditions. Another interesting observation obtained was that the mEOF of pp.TG surfaces remained almost unchanged, even at high brinogen concentrations. Since one of the limitations in the current-monitoring method is the decrease in EOF measurement sensitivity as the time required to reach plateau increases, it is difcult to ascertain if there was no brinogen adsorbed, or just very small amounts were present on the pp.TG surface. The absolute sensitivity of the experiment is therefore unclear and would in fact be expected to vary depending on the initial surface charge density. Analyses using XPS reported in our previous studies showed a good data correlation, where the amount ;of adsorbed brinogen within the channel was o10 ng/cm2 [14]. In order to evaluate the effects of different protein characteristics (in this instance both charge and size), the effects of the adsorption of chicken egg lysozyme (pI11.3) [63] on EOF in various microchannel surfaces were investigated. Results are shown in Fig. 7. On non-coated and pp.AAc-coated microchannel surfaces, a decrease in mEOF was observed when the concentration of lysozyme was increased. The adsorption of net positively charged lysozyme on net negatively charged non-coated and pp.AAc-coated surfaces reduced the effective surface charge, and thus EOF. In common with brinogen, the mEOF was unchanged on the pp.TG surface, which suggests that the adsorption of lysozyme on pp.TG surface was very low, if not eliminated by the coating. The mEOF for the pp.AAm-coated surface was observed to remain similar over the course of lysozyme concentrations tested, as shown in Fig. 7. It was, however, unclear whether lysozyme was adsorbed on the pp.AAm surface, since both lysozyme and pp.AAm surface are net positively charged at pH 7.4. Therefore, other types of surface characterization have to be carried out. Adsorption of lysozyme on a pp.AAm surface studied by ELISA (carried out as described in [14]) at 500 mg/mL and 30 min incubation had an absorbance of 1.0170.06 at 405 nm, which clearly

indicates that the protein has adsorbed to the pp.AAmcoated surface.

4 Concluding remarks
Different types of plasma-polymerized surfaces (pp.TG, pp.AAc and pp.AAm) with variable charges and charge densities were characterized using an EOF current-monitoring method as a function of buffer pH. Understanding the effects of surface chemistry on EOF, especially under different solution pH, may allow one to select the operating conditions suitable for CE optimization. All the plasma polymer surfaces exhibit slower EOF compared with noncoated microchannels. A comparison between plasma polymer and surfaces formed by wet chemistry was also carried out, where plasma polymer and covalent-grafted surfaces exhibited higher surface stabilities than physically adsorbed surfaces. The extent of protein adsorption on plasma-polymerized microchannel surfaces was shown to affect the EOF characteristics, where protein adsorption was shown to signicantly inuence electroosmotic mobility. The results highlighted that plasma polymerization was an effective method for controlling EOF behavior within microchannel. Critically, the results demonstrate that pp.TG effectively reduces protein adhesion while maintaining stable EOF, suggesting that this coating enables the development of devices capable of efcient and stable separation of proteins. The authors thank the EPSRC for funding (Grants GR/ S84347/01 and EP/E036252/1). They also thank Dr. Brian OSullivan for the microuidic chip fabrication, and Dr. Gautam Mishra for performing the XPS runs. The authors have declared no conict of interest.

5 References
[1] Strege, M. A., Lagu, A. L. (Eds.), Capillary Electrophoresis of Proteins and Peptides in Methods in Molecular Biology, vol. 276, Humana Press, 2004. [2] Camilleri, P., in: Camilleri, P. (Ed.), Capillary Electrophoresis: Theory and Practice. 2nd Edn., CRC Press, Boca Raton, FL 1998, p. 537. [3] Lee, G.-B., Lin, C.-H., Lee, K.-H., Lin, Y.-F., Electrophoresis 2005, 26, 46164624. [4] Cheng, Q. X. a.J., in: Mitchelson, K. R. (Ed.), Chip Capillary Electrophoresis and Total Genetic Analysis Systems, in New High Throughput Technologies for DNA Sequencing and Genomics, Elsevier, 2007, p. 65.

Figure 7. Electroosmotic mobilities of non-coated, pp.TG, pp.AAc and pp.AAm-coated microchannels that were incubated with lysozyme solutions (0500 mg/mL) for 30 min at pH 7.4. Each point represents an average of three independent EOF measurements.

k, V., Electrophoresis 2001, 22, [5] Horvath, J., Doln 644655. [6] Schasfoort, R. B. M., Exp. Rev. proteomics 2004, 1, 123132.

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.electrophoresis-journal.com

1886

M. Salim et al.

Electrophoresis 2009, 30, 18771887

[7] Belder, D., Ludwig, M., Electrophoresis 2003, 24, 35953606. [8] Pallandre, A., de Lambert, B., Attia, R., Jonas, A. M., Viovy, J.-L., Electrophoresis 2006, 27, 584610. [9] Ghosal, S., Annu. Rev. Fluid Mech. 2006, 38, 309338. [10] Pak Kin, W., Tza-Huei, W., Deval, J. H., Chih-Ming Ho, A., IEEE ASME Trans. Mechatron. 2004, 9, 366376. [11] Goddard, J. M., Hotchkiss, J. H., Prog. Polym. Sci. 2007, 32, 698725. [12] Hanson, K. L., Filipponi, L., Nicolau, D. V., in: Mu ller, U. R., Nicolau, D. V. (Eds.), Biomolecules and Cells on Surfaces-Fundamental Concepts, Microarray Technology and its Application, Springer, New York 2005, pp. 2344. [13] Andrade, J. D., Hlady, V., Wei, A. P., Pure Appl. Chem. 1992, 64, 17771781. [14] Salim, M., OSullivan, B., McArthur, S. L.., Wright, P. C., Lab Chip 2007, 7, 6470. [15] Lionello, A., Josserand, J., Jensen, H., Girault, H. H., Lab Chip 2005, 5, 254260. [16] Rossier, J. S., Gokulrangan, G., Girault, H. H., Svojanovsky, S., Wilson, G. S., Langmuir 2000, 16, 84898494. [17] Liu, J., Lee, M. L., Electrophoresis 2006, 27, 35333546. k, V., Electrophoresis 2004, 25, 35893601. [18] Doln [19] Popat, K. C., Desai, T. A., Biosens. Bioelectron. 2004, 19, 10371044. [20] Kadowaki, M., Yoshizawa, H., Mori, S., Suzuki, M., Thin Solid Films 2006, 506, 123127. [21] Yoshiki, H., Oki, A., Ogawa, H., Horiike, Y., Thin Solid Films 2002, 407, 156162. [22] Loughran, M., Tsai, S. W., Yokoyama, K., Karube, I., Curr. Appl. Phys. 2003, 3, 495499. [23] Tsai, S.-W., Loughran, M., Hiratsuka, A., Yano, K., Karube, I., Analyst 2003, 128, 237244. [24] Barbier, V., Tatoulian, M., Li, H., Are-Khonsari, F., Ajdari, A., Tabeling, P., Langmuir 2006, 22, 52305232. [25] Salim, M., Mishra, G., Fowler, G. J. S., OSullivan, B., Wright, P. C., McArthur, S. L., Lab Chip 2007, 7, 523525. [26] Beck, A. J., Jones, F. R., Short, R. D., Polymer 1996, 37, 55375539. [27] Hayakawa, T., Yoshinari, M., Nemoto, K., Biomaterials 2004, 25, 119127. [28] Jafari, R., Tatoulian, M., Morscheidt, W., Are-Khonsari, F., React. Funct. Polym. 2006, 66, 17571765. [29] Walker, A. K., Qiu, H., Wu, Y., Thimmons, R. B., Kinsel, G. R., Anal. Biochem., 1999, 271, 123130. [30] Yasuda, H., Plasma Polymerization. Academic Press, London 1985, p. 432. [31] Beck, A. J., Whittle, J. D., Bullett, N. A., Eves, P., Neil, S. M., McArthur, S. L., Shard, A. G., Plasma Process. Polym. 2005, 2, 641649. [32] Borkovec, M., Koper, G. J. M., Macromolecules 1997, 30, 21512158. [33] Kato, N., Schuetz, P., Fery, A., Caruso, F., Macromolecules 2002, 35, 97809787.

[34] Tyan, Y.-C., Liao, J.-D., Jong, S.-B., Liao, P.-C., Yang, M.-H., Chang, Y.-W., Klauser, R., J. Mater. Sci. Mater. Med. 2005, 16, 135142. [35] Beamson, G., Briggs, D., The ESCA300 Database, Wiley, Chichester 1992. [36] Huang, X., Gordon, M. J., Zare, R. N., Anal. Chem. 1988, 60, 18371838. [37] Schlabach, T., Weinberger, S., in: Hurst, W. J. (Ed.), Automation in Capillary Electrophoresis, Automation in the Laboratory, VCH Publishers, New York 1995, p. 23. [38] Frazier, R. A., Ames, J. M., Nursten, H. E., in: Richard, J. M. A., Frazier, A., Nursten, H. E. (Eds.), Capillary Electrophoresis for Food Analysis: Method Development, The Royal Society of Chemistry, Cambridge 2000, p. 41. [39] Lambert, W. J., Middleton, D. L., Anal. Chem. 1990, 62, 15851587. [40] Gaudioso, J., Craighead, H. G., J. Chromatogr. A 2002, 971, 249253. [41] Scales, P. J., Grieser, F., Healy, T. W., White, L. R., Chan, D. Y. C., Langmuir 1992, 8, 965974. [42] Kirby, B. J., Ernest, J., Hasselbrink, F., Electrophoresis 2004, 25, 187202. [43] Whittle, J. D., Short, R. D., Douglas, C. W. I., Davies, J., Chem. Mater. 2000, 12, 26642671. [44] Shen, M., Wagner, M. S., Castner, D. G., Ratner, B. D., Horbett, T. A., Langmuir 2003, 19, 16921699. [45] Lok, B. K., Cheng, Y.-L., Robertson, C. R., J. Colloid Interface Sci. 1983, 91, 87103. [46] Griesser, H. J., Chatelier, R. C., Gengenbach, T. G., Johnson, G., Steele, J. G., in: Cooper, S. L., Bamford, C. H., Tsuruta, T. (Eds.), Growth of Human Cells on Plasma Polymers: Putative Role of Amine and Amide Groups, in Polymer Biomaterials in Solution, as Interfaces and as Solids, VSP BV, Zeist 1995, p. 441. [47] Kolasinska, M., Warszynski, P., Bioelectrochemistry 2005, 66, 6570. [48] Oran, U., Swaraj, S., Lippitz, A., Unger, W. E. S., Plasma Process. Polym. 2006, 3, 288298. [49] Kitagawa, F., Kubota, K., Sueyoshi, K., Otsuka, K., Sci. Technol. Adv. Mater. 2006, 7, 558565. [50] Piggott, A. M., Karuso, P., Tetrahedron Lett. 2007, 48, 74527455. [51] Webster, A., Halling, M. D., Grant, D. M., Carbohydr. Res. 2007, 342, 11891201. [52] Wang, C., Wong, T. N., Yang, C., Ooi, K. T., Int. J. Heat Mass Transf. 2007, 50, 31153121. [53] Issaq, H., Atamna, I., Muschik, G., Janini, G., Chromatographia 1991, 32, 155161. [54] Tang, G. Y., Yan, D. G., Yang, C., Gong, H. Q., Chai, C. J., Lam, Y. C., J. Phys. Conf. 2006, 34, 925930. [55] Fane, J. K., Nystrom, A. G., Pihlajamaki, M., Bowen, A., Mukhtar, W. R. K., J. Membr. Sci. 1996, 116, 149159. [56] Szymczyk, A., Fievet, P., Mullet, M., Reggiani, J. C., Pagetti, J., J. Membr. Sci. 1998, 143, 189195.

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.electrophoresis-journal.com

Electrophoresis 2009, 30, 18771887

Microuidics and Miniaturization

1887

[57] Vernhet, A., Bellon-Fontaine, M. N., Doren, A., J. Chim. Phys. 1994, 91, 17281747. [58] Werner, C., Kro ber, H., Zimmermann, R., Dukhin, S., Jacobasch, H.-J., J. Colloid Interface Sci. 1998, 208, 329346. [59] Erickson, D., Li, D., Werner, C., J. Colloid Interface Sci. 2000, 232, 186197.

[60] Sze, A., Erickson, D., Ren, L., Li, D., J. Colloid Interface Sci. 2003, 261, 402410. [61] Ghosal, S., Anal. Chem. 2002, 74, 771775. [62] Yeung, K. K. C., Lucy, C. A., Anal. Chem. 1997, 69, 34353441. [63] Wetter, L. R., Deutsch, H. F., J. Biol. Chem. 1951, 192, 237242.

& 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.electrophoresis-journal.com