Time-of-Flight Secondary Ion Mass Spectrometry Study of the

Nov 2, 2008 - Repair and Regeneration, Institute of Health and Biomedical InnoVation and ... Molecular and Microbial Sciences and Australian Institute...
0 downloads 0 Views 726KB Size
Download PDF
Langmuir 2009, 25, 1011-1019

1011

Time-of-Flight Secondary Ion Mass Spectrometry Study of the Orientation of a Bifunctional Diblock Copolymer Attached to a Solid Substrate Marek Jasieniak,† Shuko Suzuki,‡,§ Michael Monteiro,| Edeline Wentrup-Byrne,‡ Hans J. Griesser,*,† and Lisbeth Grøndahl*,§ Ian Wark Research Institute, UniVersity of South Australia, Mawson Lakes, South Australia 5095, Tissue Repair and Regeneration, Institute of Health and Biomedical InnoVation and School of Physical and Chemical Sciences, Queensland UniVersity of Technology, Brisbane, Queensland 4001, and School of Molecular and Microbial Sciences and Australian Institute for Bioengineering and Nanotechnology, The UniVersity of Queensland, Brisbane, Queensland 4072, Australia ReceiVed June 26, 2008. ReVised Manuscript ReceiVed NoVember 2, 2008 A block copolymer consisting of a phosphate-containing moiety (poly[2-(methacryloyloxy)ethyl phosphate], PMOEP) and a keto-containing moiety (poly[2-(acetoacetoxy)ethyl methacrylate], PAAEMA) showed good stability after attachment to an APS amine-modified glass slide, as did both of the respective homopolymers. The PAAEMA homopolymer can attach to the APS amine groups via covalent linkages, while the PMOEP homopolymer most likely attaches through electrostatic interactions involving deprotonated phosphate and protonated amine groups. To elucidate the conformation of the block copolymer after attachment, particularly with respect to the PMOEP segment orientation, principal component analysis (PCA) of time-of-flight secondary ion mass spectrometry (ToF-SIMS) spectra of the surface-attached polymer layers was performed. Comparison with the pure homopolymer spectra and interpretation after PCA indicate that the adsorbed conformation is not random. Rather, the copolymer is adsorbed in a conformation that preferentially exposes the PMOEP block toward the outer surface. We thus conclude that the most likely conformation of PMOEP-b-PAAEMA immobilized onto the APS-modified glass slide is via covalent interfacial linkages involving the PAAEMA block with the result that the surface is enriched in PMOEP tails. This in turn implies that under the conditions applied (dry DMF) the covalent coupling of keto groups to the amine groups of the aminated slide is more efficient than the proton transfer required for the generation of electrostatic attractions. This (partially) preferential orientation of the PMOEP-b-PAAEMA copolymer could have significant implications on interfacial interactions such as those involved in nucleation and the subsequent mineralization sequence of events in hydroxyapatite formation. The present study demonstrates that ToF-SIMS is a powerful tool not only for the investigation of the surface composition of adsorbed layers, but also for probing the molecular conformation of such adsorbed block copolymers, though care is required in the PCA analysis of multiple spectra.

Introduction Tailored block copolymers with well-defined structures, composition, and properties can be produced using living free radical polymerization methods such as reversible addition fragmentation chain transfer (RAFT).1-4 A homopolymer with tunable chain length produced by the RAFT technique still contains the RAFT end group and can therefore be used as a macroinitiator for the production of various block copolymers.5-8 * To whom correspondence should be addressed. Phone: + 61 7 3365 3671 (L.G.); + 61 8 8302 3703 (H.J.G.). Fax: + 61 7 3365 4299 (L.G.); + 61 8 8302 3683 (H.J.G.). E-mail: [email protected] (L.G.); [email protected] (H.J.G.). † University of South Australia. ‡ Queensland University of Technology. § School of Molecular and Microbial Sciences, The University of Queensland. | Australian Institute for Bioengineering and Nanotechnology, The University of Queensland. (1) Moad, G.; Rizzardo, E.; Thang, S. H. Aust. J. Chem. 2005, 58, 379–410. (2) Chong, B. Y. K.; Le, T. P. T.; Moad, G.; Rizzardo, E.; Thang, S. H. Macromolecules 1999, 32, 2071–2074. (3) Perrier, S.; Takolpuckdee, P. J. Polym. Sci., Part A: Polym. Chem. 2005, 43, 5347–5393. (4) Monteiro, M. J. J. Polym. Sci., Part A: Polym. Chem. 2005, 43, 3189– 3204. (5) Chong, Y. K.; Krstina, J.; Le, T. P. T.; Moad, G.; Postma, A.; Rizzardo, E.; Thang, S. H. Macromolecules 2003, 36, 2256–2272. (6) Mao, M.; Turner, S. R. J. Am. Chem. Soc. 2007, 129, 3832–3833. (7) Ge, Z.; Xie, D.; Chen, D.; Jiang, X.; Zhang, Y.; Liu, H.; Liu, H. Macromolecules 2007, 40, 3538–3546.

Numerous studies of block copolymers produced by RAFT polymerization exist in the literature. It has been shown that micelle formation by block copolymers can be controlled, not only by the chemical composition, side or chain-end functionalities, and chain length but also by the environment (solvent, pH, and temperature).9,10 Lithographic applications have utilized this ability of block copolymers to spontaneously assemble into ordered structures such as spheres, cylinders, and lamellae by enabling their deposition onto chemically heterogeneous surfaces.11 In this type of application the block copolymers are typically deposited onto a surface by spin casting followed by annealing. Polymers can also be covalently attached to a substrate by first anchoring RAFT agents to the surface followed by either a “grafting-from” or “grafting-to” polymerization to produce a coated surface.12 (8) Cheng, C.; Schmidt, M.; Zhang, A.; Schlu¨ter, A. D. Macromolecules 2007, 40, 220–227. (9) Schilli, C. M.; Xhang, M.; Rizzardo, E.; Thang, S.h.; Chong, Y. K.; Edwards, K.; Karlsson, G.; Mu¨ller, A. H. E. Macromolecules 2004, 37, 7861–7866. (10) Zhang, L.; Nguyen, T. L. U.; Bernard, J.; Davis, T. P.; Barner-Kowollik, C.; Stenzel, M. H. Biomacromolecules 2007, 8, 2890–2901. (11) Park, M.; Harrison, C.; Chaikin, P. M.; Register, R. A.; Adamson, D. H. Science 1997, 276, 1401–1404. (12) Peng, Q.; Lai, D. M. Y.; Kang, E. T.; Neoh, K. G. Macromolecules 2006, 39, 5577–5582. (13) Suzuki, S.; Whittaker, M. R.; Grøndahl, L.; Monteiro, M. J.; WentrupByrne, E. Biomacromolecules 2006, 7, 3178–3187. (14) Chirila, T. V.; Zainuddin, React. Funct. Polym. 2007, 67, 165–172.

10.1021/la802016b CCC: $40.75  2009 American Chemical Society Published on Web 12/24/2008

1012 Langmuir, Vol. 25, No. 2, 2009 Scheme 1. Structure of the PMOEP-b-PAAEMA Copolymer (n ) 108, m ) 97)

Jasieniak et al. Scheme 2. Schematic Depictions of Possible Orientations of the Block Copolymer and Chemical Nature of Attachment for PMOEP-b-PAAEMA Functionalized on an APS Slidea

a Key: (a) attachment of both copolymers, (b) attachment of PAAEMA, (c) attachment of PMOEP. The thick blue line represents PAAEMA, and the thin purple line represents PMOEP.

We recently reported the first RAFT-mediated polymerization of soluble poly[2-(methacryloyloxy)ethyl phosphateb-2-(acetoacetoxy)ethyl methacrylate] (PMOEP-b-PAAEMA)) diblock copolymers, Scheme 1.13 The polymers were designed such that the reactive keto groups of the PAAEMA block could be covalently attached to an aminated surface while the PMOEP block was incorporated to stimulate calcium phosphate nucleation. Mineralization is a desired property of biomaterials intended for use at a bone interface.14,15 When the PMOEP-b-PAAEMA diblock copolymer was subsequently attached to aminated glass slides for mineralization studies, it showed an unexpectedly poor mineralization outcome.13 A key question concerned the orientation of the diblock copolymer at the solid surface. Would it adopt a random conformation or a (partially) segregated conformation with a phosphate block at the copolymer/water interface, or would the phosphate groups preferentially orient toward the substrate surface to maximize ionic interfacial bonding with amine groups on the substrate surface? In initial attempts to characterize the polymer adlayer we used X-ray photoelectron spectroscopy (XPS). XPS not only enables the quantitative determination of the amounts of adsorbed polymer adlayers present, but also serves to characterize the aminated slide substrates and interfacial reactions. The N 1s high-resolution scans proved particularly useful in the present case because they provided information about the relative amounts of the different nitrogen species present such as amines, protonated amines, imines, and amides.16-18 However, since the probe depth of XPS is too large to elucidate conformational orientations of polymers of the molecular weights studied, we exploited the complementary capabilities of static time-of-flight secondary ion mass spectrometry (ToF-SIMS) in this study. Static ToF-SIMS has a much higher surface sensitivity than XPS and has previously been utilized to study the conformation/ orientation of surface-adsorbed protein adlayers.19,20 Knowledge of the protein amino acid sequence and the native conformation was combined with an analysis of the amino acid mass fragments detected in the static ToF-SIMS experiment to obtain the relevant conformational information. Due to the complexity of the mass spectral data, the application of multivariate analysis methods such as principal component analysis (PCA) is often used to refine interpretation of the experimental data. This approach makes it possible to reduce (15) Kamei, S.; Tomita, N.; Tamai, S.; Kato, K.; Ikada, Y. J. Biomed. Mater. Res. 1997, 37, 384–393. (16) Gengenbach, T. R.; Chatelier, R. C.; Griesser, H. J. Surf. Interface Anal. 1996, 24, 611–619. (17) Tyler, B. J.; Castner, D. G.; Ratner, B. G. Surf. Interface Anal. 1989, 14, 443–450. (18) Lawrie, G. A.; Grøndahl, L.; Battersby, B. J.; Keen, I.; Lorentzen, M.; Surawski, P.; Trau, M. Langmuir 2006, 22, 497–505. (19) Wagner, M. S.; Castner, D. G. Appl. Surf. Sci. 2004, 231-232, 366–376. (20) Griesser, H. J.; McArthur, S. L.; Wagner, M. S.; Kingshott, P.; McLean, K. M. In Biopolymers at Interfaces, 2nd ed.; Malmsten, M., Ed.; Marcel Dekker Inc.: New York, Basel, Switzerland, 2003.

large multidimensional data sets to a few plots that describe major variations in the data.21 Static ToF-SIMS has proven a most useful technique for the evaluation of the conformations of a range of macromolecules on surfaces based on a significant variation in the relative intensity of specific peaks between related samples. In one such study the relative intensities of peaks in the positive ToF-SIMS spectra were used to evaluate the orientation of phospholipids assembled on silica.22 In another study, it was reported that the different tacticities of poly(methyl methacrylate-b-styrene) block copolymers exhibited significant differences in relative peak intensities.23 A ToF-SIMS investigation of silicon-grafted diblock copolymers showed exclusively the presence of peaks from the block at the air interface, in agreement with the estimated thickness of this block copolymer layer being 10 nm.12 In this study, we combine ToF-SIMS with PCA to study the orientation of the PMOEP-b-PAAEMA diblock copolymer (Scheme 1) attached to a solid substrate. The results, in conjunction with the XPS data and after comparison with two control samples consisting of adlayers of the pure homopolymers PAAEMA and PMOEP, led us to conclude that the block copolymer is attached in a partially oriented manner, with a degree of enrichment of phosphate groups toward the outer surface. In addition to the demonstration that ToF-SIMS can determine the configuration of attached diblock copolymers, the information generated in this study is also relevant to the interpretation of the performance of this type of adlayers, such as for biomineralization studies.

Materials and Methods Fabrication of the Polymer Adlayer. The homopolymers and the block copolymer used in this study were synthesized using RAFT techniques as previously described.13 (3-Aminopropyl)trimethoxysilane (APS)-treated glass slides were kindly donated by Asper Biotech (ES). The functionalization of the APS slides has also been described previously.13 In summary, the reaction of PAAEMA and PMOEP-b-PAAEMA with the APS slide was carried out by immersing the slide in a polymer solution in dry DMF followed by the addition of NaCNBH3. The intention was to form a covalent interfacial linkage by reductive amination between the keto group of the PAAEMA side chains and the surface amines, thus creating interfacial secondary amine bonds. Another possibility is that the copolymer might also adsorb onto the amine surfaces via electrostatic interactions between phosphates on the PMOEP block and the surface amine groups. Similarly, the APS slide was immersed in a PMOEP solution in dry DMF; PMOEP presumably adheres via electrostatic interactions. Subsequently, all slides were washed thoroughly with DMF, rinsed with acetone, and dried. The good stability of the (21) Graham, D. J.; Wagner, M. S.; Castner, D. G. Appl. Surf. Sci. 2006, 252, 6575–6581. (22) Pacholski, M. L.; Cannon, D. M.; Ewing, A. G.; Winograd, N. J. Am. Chem. Soc. 1999, 121, 4716–4717. (23) Eyende, X. V.; Weng, L. T.; Bertrand, P. Surf. Interface Anal. 1997, 25, 41–45.

Study of the Orientation of an Adsorbed Copolymer

Figure 1. Positive static SIMS spectra for (a) APS, (b) APS-PAAEMA, (c) APS-PMOEP, and (d) APS-PMOEP-b-PAAEMA.

functionalized slides in aqueous solution was confirmed by immersing the slides in simulated body fluid24 (SBF) for 1 week at 36.5 °C and comparing XPS data obtained for these slides with those obtained for freshly prepared slides.13 X-ray Photoelectron Spectroscopy. XPS spectra were recorded using a Kratos Axis Ultra X-ray photoelectron spectrometer with a monochromated Al KR X-ray source (1486.6 eV) running at 150 W (15 kV, 10 mA emission current). The survey scans were collected at 1200-0 eV with 1.0 eV steps at a pass energy of 160 eV and the narrow scans at 0.1 eV steps at a pass energy of 20 eV. Vision 2 software was used for data acquisition and processing. The binding energies were charge-corrected using the C 1s peak (285 eV). Static Time-of-Flight Secondary Ion Mass Spectrometry. The ToF-SIMS analyses were performed with a PHI TRIFT II (model 2100) spectrometer (PHI Electronics Ltd.) equipped with a 69Ga liquid metal ion gun (LMIG). A 15 keV pulsed primary ion beam was used to desorb and ionize species from a sample surface. Pulsed, low-energy electrons were used for charge compensation. Stainless steel grids were additionally used to minimize charging effects. Mass axis calibration was done with CH3+, C2H5+, and C3H7+ in positive mode and with CH-, C2H-, and Cl- in negative mode of operation. A mass resolution m/∆m of ∼4500 at nominal m/z ) 27 amu (C2H3+) was typically achieved. Although this technique is “destructive” by its very nature, by applying an ion beam of low current, it is possible to derive data from a virtually intact surface. In this study, the primary ion fluxes used were between 3 × 1011 and 6 × 1011 ions cm-2, meeting the static conditions regime.25 This means that, for the time period of the measurement, less than 0.1% of the surface atomic sites (1 in 1000) are struck by the primary ion beam. The spectra recorded under these “static SIMS conditions” provide mass spectrometric information from surface layers with a negligible contribution from multiple ion impacts. Each sample was characterized by 8 positive and 5 negative spectra: 49 positive and 49 negative peaks were selected for the sample characterizations (Table 2). It was ascertained that this peak selection (24) Kokubo, T.; Takadama, H. Biomaterials 2006, 27, 2907–2915. (25) Briggs, D. Surface Analysis of Polymers by XPS and Static SIMS; Cambridge University Press: Cambridge, U.K., 1998.

Langmuir, Vol. 25, No. 2, 2009 1013

Figure 2. Negative static SIMS spectra for (a) APS, (b) APS-PAAEMA, (c) APS-PMOEP, and (d) APS-PMOEP-b-PAAEMA.

did not introduce arbitrary errors in interpretation by alternative selections not shown. The complex ToF-SIMS sets of data were evaluated with the aid of PCA and analysis of means.26,27 The original data matrix was built of n columns comprising the intensities of peaks selected from the mass spectra and m rows corresponding to m spectra subjected to multivariate analysis. The peak intensities were normalized to the total intensity of the selected peaks and mean-centered prior to PCA. The process of data decomposition by PCA results in the formation of two new matrices: a scores matrix and a loadings matrix. The scores show the relationship between the samples in the new coordinate (PC-defined) space, while the loadings illustrate the relationship between the original variables and the principal components. Principal component analysis was accomplished using PLS_ Toolbox version 3.0 (Eigenvector Research, Inc., Manson, WA) working in the MATLAB platform (MATLAB version 6.5, The MathWorks Inc., Natic, MA).

Results and Discussion Preparation of Polymer Adlayer Samples. In the present study, APS-modified glass slides were used as the substrates for separate immobilization of the two homopolymers PAAEMA (units 23) and PMOEP (units 21) as well as the PMOEP (units 108)-PAAEMA (units 97) block copolymer (PMOEP-bPAAEMA, Scheme 1), each forming polymer adlayers on the aminated glass substrate. These polymers were prepared by RAFT-mediated polymerization as described previously.13 Coupling of the polymers containing keto groups, PAAEMA and PMOEP-b-PAAEMA, to the APS slides was carried out in dry DMF at room temperature followed by mild reduction with NaCNBH3 to convert the unstable interfacial imine linkages to stable secondary amines. PMOEP was immobilized onto the APS slide by immersion in a solution of PMOEP in dry DMF. This aprotic solvent is not expected to cause deprotonation of the phosphate groups of the PMOEP moieties; however, it is (26) Zuwaylif, F. H. General Applied Statistics; Addison-Wesley Publishing Co., Inc.: Reading, MA, 1980. (27) Miller, J. C.; Miller, J. N. Statistics for Analytical Chemistry; Ellis Horwood Ltd.: Chichester, U.K., 1993.

1014 Langmuir, Vol. 25, No. 2, 2009

Jasieniak et al.

Table 1. XPS Data: Element Concentrations (atom %) from XPS Survey Scans sample

N

C

P

APS APS-PAAEMA APS-PMOEP APS-PMOEP-b-PAAEMA

2.6 1.8 1.7 1.5

22.3 26.3 26.1 36.0

1.3 1.5

likely that proton transfer between the phosphate groups of PMOEP and the amine groups of APS can occur, thus achieving electrostatically driven adsorption. In the case of the block copolymer it was envisaged that a competition between reductive amination (leading to covalent attachment of the PAAEMA block) and proton transfer (leading to electrostatic attachment of the PMOEP block) might exist. The final conformation of the block copolymer will be decided by the relative efficiencies of these two possible interfacial immobilization mechanisms. We considered three different configurations (Scheme 2), one in which the PAAEMA block is covalently attached (b), one in which the PMOEP block is attached through ionic interactions (c), and one combining both types of attachment (a). These three configurations provide a basis for the qualitative prediction of theoretical spectra against which the measured spectra can be compared. XPS Characterization of Polymer Adlayers. The successful immobilization of the polymers onto the APS-coated slides was verified using solution stability studies and from XPS data, as described previously.13 The XPS spectrum of the APS slide used in the current study showed the presence of N and C from the APS (Table 1), as well as O and Si and small amounts of Na and Ca from the glass slide. After reaction with PMOEP-b-PAAEMA, the amount of C increased (22-36 atom %) and a signal assignable to P appeared, indicating successful attachment of the block copolymer. Immobilization of the homopolymers PAAEMA and PMOEP onto the APS slide resulted in a reduction in the amount of N and an increase in the amount of C (to 26 atom % in both systems), confirming successful attachment of the homopolymers. This indicates that a higher amount was attached for the block copolymer than for the homopolymers. In turn, this suggests that the block copolymer attachment is more complex than a simultaneous attachment of the individual blocks; i.e., Scheme 2a is an unlikely scenario for the block copolymer. In addition, the PMOEP-modified slide showed the appearance of P as expected, the amount being a little less than that observed for the block copolymer. The stability of the modified slides was established by soaking them in SBF for seven days. All surfaces were found to be stable for this period as evaluated by XPS.13 First, this suggests that successful reductive amination occurred upon formation of the PAAEMA adlayer since this uncharged polymer (Mn of only ∼5000) should be less stably adsorbed if no covalent interfacial bonds were formed. Second, it indicates that strong polyelectrolyte interactions are indeed forming, at least in the PMOEP homopolymer adlayer, due to the deprotonated PMOEP. This homopolymer data set indicates that the block copolymer could attach to the amine surface by either mechanism. Since it cannot be predicted which would be the dominant mechanism, the nature of the chemical interaction of the block copolymer with the APS surface was further evaluated by investigating the high-resolution XPS N 1s spectra. The high-resolution spectra of the N 1s region of all samples required three components for the curve fit: amine (399.6 eV), amide (401 eV), and protonated amine (402 eV).13 No oxidized nitrogen species (such as NO3-) were detected by XPS at BE values >402 eV for any of the samples. The presence of amide

Figure 3. Differences between samples in high-mass-resolution mode (m/z ≈ 42 amu): (a) APS, (b) APS-PAAEMA, (c) APS-PMOEP, (d) APS-PMOEP-b-PAAEMA.

groups in all the samples indicates partial oxidation of the APS moiety. The amount of amide groups was similar for all samples (23-27% of total nitrogen), suggesting that these amides are inherent in the APS layer and that no further oxidation resulted from the procedures used to attach the polymers to the slides. The relative ratios of the fitted amine and protonated amine peaks revealed that, compared to the APS slide (amine:protonated amine ) 46:26), the attachment of PAAEMA caused a slight decrease in the protonated amine contribution (55:22). This is in agreement with the coupling of this homopolymer through reductive amination (Scheme 2b). The attachment of PMOEP caused a large increase in the protonated amine peak (31:43), which provided evidence for ionic interactions between the deprotonated phosphate groups and the protonated amine groups of the APS slide (Scheme 2c). When the APS-PMOEP-b-PAAEMA block copolymer sample was compared to the APS slide, an increase in the component assigned to the protonated amine was observed (to a ratio of 42:32). However, this increase was not as large as for the PMOEP-attached slide. Previously this led us to the conclusion that both reductive amination and ionic interactions might be involved in block copolymer attachment as depicted in Scheme 2a.13 However, the protonation of amine groups might also occur if the block copolymer is attached via reductive amination. This would be facilitated by proton transfer from those phosphate groups unable to get close enough to the surface to undergo ionic bonding. The following question now arises: What is the relative importance of the two interfacial bonding mechanisms? Characterization of the Polymer Adlayers by ToF-SIMS. The APS-coated slide as well as PAAEMA-, PMOEP-, and PMOEP-b-PAAEMA-functionalized slides were subjected to ToF-SIMS analysis. The phosphate groups of the PMOEP moiety are easily recognizable by ToF-SIMS in the negative ion spectra and, consequently, are very suitable for conformational studies. The positive static SIMS spectrum of the APS slide (Figure 1a) shows distinct hydrocarbon peaks. The signal at m/z ) 28 contains two close-lying peaks corresponding to Si+ and C2H4+. The peak intensities were quite consistent across different measurement spots on a sample, indicating good

Study of the Orientation of an Adsorbed Copolymer

Langmuir, Vol. 25, No. 2, 2009 1015

Table 2. Positive and Negative Fragments Used in PCA positive fragments +

+

+

+

+

+

negative fragments +

+

+

C , CH , CH2 , CH3 , C2H2 , C2H3 , C2H4 , C2H5 , C2H6 , C3H2+, C3H3+, C3H4+, C3H5+, C3H6+, C3H7+, C3H8+, C4H3+, C4H4+, C4H5+, C4H6+, C4H7+, C4H8+, C4H9+, C4H10+, C5H5+, C5H6+, C5H7+, C5H8+, C5H9+, C5H510+, C5H11+, Al+, Si+, SiH+, CH2Si+, CH3Si+, SiOH+, CHO+, CH3O+, C2H2O+, C2H3O+, C2H4O+, COOH+, C2H5O+, C4H5O+, CH2NO+, CH3NO+,CH4N+, C2H4N+

uniformity of the APS layer. The spectrum also reveals the presence of sodium. The negative ion spectrum of the APS slide is dominated by signals at m/z ) 16 (O-) and at m/z ) 17 (OH-). Signals at m/z ) 46 (NO2-) and at m/z ) 62 (NO3-) (Figure 2a) are of relatively high intensities despite the fact that no oxidized nitrogen species were observed by XPS. This suggests a low surface concentration of these species and their over-representation in the negative mass spectrum by a high sputter yield. The positive mass spectra of the APS slides both before and after modification with PAAEMA, PMOEP, and PMOEP-bPAAEMA are shown in Figure 1. In these vertically compressed displays, only the most intense peaks are clearly visible, but between them, many other peaks are observable with adequate signal-to-noise ratios. Despite similar qualitative characteristics, the intensity patterns vary significantly between the samples. For example, spectra a and d indicate a substantial reduction in the yield of the Si+ ion (m/z ) 28) after immobilization of PMOEP-b-PAAEMA which is assignable to reduced substrate ion emission due to the presence of an adlayer. We do not expect these polymer adlayers to achieve full surface coverage as they dry in vacuum during sample preparation for ToF-SIMS analysis, hence the continued presence of substrate ion signals. The negative mass spectra (Figure 2) clearly show the emergence of peaks at m/z ) 63 and m/z ) 79 upon immobilization of PMOEP and PMOEP-b-PAAEMA. These peaks correspond to the PO2- and PO3- fragment ions and are indicative of the phosphoruscontaining moieties on the surface. This is in agreement with the fact that, in XPS, P is only observed in these samples. Further information can be obtained by high-resolution analysis of regions that contain close-lying peaks, for example, the peak at m/z ) 42 seen in Figure 1 for all samples. Closer inspection reveals that this peak consists of four components whose relative intensities vary between samples (Figure 3). As expected the CH2Si+ signal assignable to aminosilanized glass decreases in intensity as the adlayers are attached. The C2H2O+ signal is of very low intensity for the APS surface as expected for a properly cured APS layer and increases in intensity for the modified slides. Interestingly, the intensity of the C2H4N+ signal varies irregularly; the efficiency of generating this ion is expected to vary not only with the amount of overlying polymer but also with the degree of ionization of the amine groups. As shown above by XPS, the presence of the three different adlayers affects the extent of protonation of the surface amine groups. Thus, it is impossible to utilize the intensity of this peak to infer contributions of specific structures. It is also clear from Figures 1 and 3, as well as from inspection of other high-resolution mass regions not shown, that the spectrum of the attached block copolymer is not simply a linear addition of the spectra of the two homopolymers. This suggests that the surface-attached conformation is not the flat, nonoriented conformation as shown in Scheme 2a. However, the spectral differences are, apart from the phosphate peaks, not unique peaks; rather, they consist of changes in relative intensities. Thus, a more detailed analysis needs to invoke a considerable number of peaks.

-

-

C , CH , CH2 , CH3 , C2-, C2H-, C2H3-, C3-, C3H-, C3H2-, C3H3-, C3H4-, C4-, C4H-, C4H2-, C4H3-, C5H-, O-, OH-, Si-, SiH-, CHSi-, SiO-, SiOH-, SiO2-, SiO2H-, SiO3-, SiO3H-, P-, CP-, PO-, POH-, PO2-, PO3-, NO-, NO2-, NO3-, CN-, CHN-, C2HN-, CNO-, CHNO-, CH3O-, C3O-, C2HO-, C2H3O-, COOH-, C3H3O-, C3H4O-

-

The univariate analysis of individual peaks described so far and their intensities across the four samples yield some structural information, but the analysis process is tedious and time-consuming, and could be ambiguous if only a small number of peaks are selected arbitrarily. Univariate analysis is warranted to separate the APS surface from the polymer adlayers as the spectra recorded on the APS surface are readily distinguishable; for example, APS cannot produce ions with more than three connected C atoms, whereas the adsorbed polymers produce larger hydrocarbon fragment ions from the polymer backbone. This difference is clearly evident in the spectra (though not readily seen in the compressed spectra of Figure 1). However, while the APS surface is readily separated spectrally from the other surfaces, a protocol more reliable than inspection of selected peaks needs to be employed to analyze the spectra of the three polymer adlayers and their conformations. Global multivariate analysis is much more reliable for the detection of statistically significant changes in the spectra. Furthermore, the loadings plots generated by PCA can be used to check for consistency between the peaks found to be most significant by PCA and the known chemical structures. Principal Component Analysis of ToF-SIMS Data. Data compression and information extraction were performed with the aid of PCA to extract the most essential and useful information. PCA was independently performed on 49 positive and 49 negative peaks (Table 2), resulting in the formation of two new matrices: a scores matrix and a loadings matrix. The relationship between the original variables (peak intensities) and the principal components is illustrated by loadings plots. The peaks that have the highest loadings within the data set have the largest influence on the separation as seen in the scores plots. The scores and the relevant loadings plots should be evaluated together, following the principle that peaks with positive loadings are relatively more intense in spectra of samples with positive scores and relatively less intense in spectra of samples with negative scores on a given PC (and vice versa).19 Although PCA is usually performed only on positive ion spectra because negative ion spectra generally contain far fewer peaks and thus much less information, for the present purpose it is essential to perform PCA analysis and discuss the loadings plots for both positive and negative ion spectra. This is particularly significant in this case due to the fact that a key molecular structure of interest, the phosphate group, appears strongly in the negative ion spectra but not in the positive ion spectra. Figures 4 and 5 show the scores for the positive and negative ion mass spectra of the polymer adlayer surfaces; the APS data are not included in the following PCA analyses as they are clearly different and not instructive for the purpose of elucidating the copolymer orientation. In both cases, there are three distinct clusters corresponding to the three samples investigated. Each cluster represents multiple spectra taken from different areas of the samples. The first two PCs captured 98% of the data variance present in both the positive and the negative mass spectra. This indicates that most of the variance within the original spectral data has been retained.

1016 Langmuir, Vol. 25, No. 2, 2009

Figure 4. Scores plot on PC1 and PC2 (a) and loadings on PC1 (b) of positive ion mass signals for APS-PAAEMA, APS-PMOEP, and APS-PMOEP-b-PAAEMA (∧ denotes hydrocarbon fragment ions). The ellipses in (a) are for visual guidance only.

Figure 5. Scores plot on PC1 and PC2 of negative ion mass spectra for APS-PAAEMA, APS-PMOEP, and APS-PMOEP-b-PAAEMA. The ellipses are for visual guidance only.

The scores plots clearly illustrate that the spectrum of the block copolymer sample is not simply a linear addition of the spectra of the two homopolymers (in which case the copolymer data should lie between the homopolymer data). This observation suggests that the attached copolymer does not adopt a fully random conformation (Scheme 2a); the fact that the PMOEP-b-PAAEMA adlayer is distinct from both homopolymers also suggests that in the block copolymer there exists a mixed surface; neither block fully covers the other. To elucidate which conformation is adopted, albeit probably only partially, analysis is required to identify which ion signals show intensities higher or lower than expected for a simple addition of the homopolymer signals. PCA is much more reliable for this than univariate analysis. In the positive spectra, the first principal component captures 96% of the spectral information and clearly separates the spectrum of the block copolymer adlayer from those of the two homopolymers, as shown by their different locations along the PC1

Jasieniak et al.

Figure 6. Loadings on PC1 (a) and PC2 (b) of negative ion mass signals for APS-PAAEMA, APS-PMOEP, and APS-PMOEP-b-PAAEMA.

axis (Figure 4a). The data clusters for the two homopolymer adlayers, however, partially overlap along the PC1 axis, and hence, PC1 alone does not reliably separate their spectral information. On the other hand, the second principal component separates the two homopolymer adlayers, although the difference is small. The location of the copolymer data, however, overlaps with those of the homopolymers on the PC2 axis. This indicates that both PC1 and PC2 should be used for full interpretation of the positive SIMS spectra of these three samples. For the present purposes, however, we are not so much concerned with differences between the two homopolymer adlayers, but with how the copolymer adlayer differs spectrally from them. Noting that the copolymer is separated from the two homopolymers along the PC1 axis, we can use the loadings on PC1 (Figure 4b) to assess the spectral differences. Positive loadings, that is, those relatively more intense for the homopolymers, are revealed for peaks assignable to the underlying layers of Si and APS. This observation is consistent with the smaller adsorbed amounts (relative to the copolymer) indicated by XPS. The peaks that load negatively and thus are relatively more intense in the copolymer spectrum are assignable to CxHyO+, such as C2H3O+ (m/z ) 43), and hydrocarbon fragment ions; this observation likewise can be attributed to higher surface coverage. No phosphorus-containing peaks were found in the positive ion data, which is most likely a function of the ionization process. For PC2, the differences between samples are similar to the experimental uncertainty and are thus insufficient for interpretation. The loadings plots contain no evidence of any surface contaminants that would confound analysis. PCA analysis of the negative ion spectra gives a better separation of the three surfaces (Figure 5). The PC1 loadings of the negative ion spectra (Figure 6a) show that the substrate signals, containing Si and NOx, are least intense for the copolymer adlayer, which again is consistent with the inference from the XPS data that this polymer adsorbs to the greatest amount. Even more

Study of the Orientation of an Adsorbed Copolymer

informative, however, is the strong negative loading of the phosphate-derived peaks. Interestingly, the copolymer sample shows a relatively higher intensity of the phosphate ions compared with the PMOEP homopolymer sample. We also note the signals assignable to NOx ions, which decrease in intensity as polymer coverage increases. These are assignable to the underlying APS layer. The presence of oxidized N fragments indicates that oxidation of the APS surface is occurring to some extent. The fact that no such signals could be seen in the XPS N 1s spectra is indicative of the much higher sensitivity of ToF-SIMS. The loadings on PC2 are positive for Si substrate signal and negative for POx and NOx species (Figure 6b). The absence of signals assignable to poly(dimethylsiloxane) (PDMS) in positive mass spectra (e.g., at m/z 73, 147, 207, 221) confirms that the Si signal is from the substrate rather than contamination. However, in contrast to PC1 as discussed above, in PC2 the copolymer adlayer shows a relatively lower intensity of the phosphate ions compared with the PMOEP homopolymer adlayer. This and the diverging trends of the substrate signals suggest that the analysis of Figure 6b may be affected by the analysis procedure. We note that the copolymer and PAAEMA adlayer samples possess the same PC2 score despite obvious chemical differences; this coincidence may affect the loadings plot. This suggests that simultaneous PCA of three (or more) samples can in some instances lead to difficulties in unambiguous interpretation due to coincidental colocation on some of the PCs. Here we have used so far the spectra of all three adlayers because we want to elucidate how APS-PMOEP-b-PAAEMA differs from the other two, but it appears that this three-way comparison leads to confounding scores and loadings. While APS-PMOEP and the copolymer adlayer are separated by both PC1 and PC2, in the present case it is difficult to ascertain with confidence an interpretation of the differences between the two adlayers from the two loadings plots. We surmise that this is due to the confounding presence of the APS-PAAEMA data. To overcome this ambiguity, we undertook a pairwise comparison between APS-PMOEP and APS-PMOEP-b-PAAEMA as well. The pairwise comparison of these two polymer adlayers provides, as expected, two clusters of data points separated along the PC1 axis (Figure 7a), whereas the PC2 axis on the other hand arises from variability within the data clusters. For APS-PMOEPb-PAAEMA most data are virtually superimposed except for one outlier. For APS-PMOEP the data fall into three clusters. The scatter within each set of spectra is, however, significantly less than the difference between the means of the clusters for the two adlayers. Thus, loadings on PC1 provide meaningful information on the spectral differences that contribute most to the separation. The loadings of the P-containing fragment ions (Figure 7b) are all negative, indicating that they are relatively more abundant in the spectra of APS-PMOEP-b-PAAEMA. Peaks containing Si and NOx, on the other hand, show positive loadings (not shown), which is consistent with the thinner adlayer of PMOEP. The scores and loadings plots show that there are significant, nonadditive differences between the spectra, both positive and negative, of the three attached polymer layers. The question now is whether we can use differences in the spectra of the adsorbed polymer layers to draw inferences about the molecular conformation of the adsorbed diblock copolymer. A polymer adlayer typically possesses regions known as loops, trains, and tails as illustrated in Scheme 2, unless strong adsorption forces force a tightly adsorbed flat conformation. In orientation b the PMOEP block is predominately oriented away from the surface as tails, while the covalently immobilized PAAEMA forms loops and

Langmuir, Vol. 25, No. 2, 2009 1017

Figure 7. (a) Scores plot of negative mass spectra and (b) loadings of P-containing fragments on PC1 for APS-PMOEP and APS-PMOEPb-PAAEMA.

trains. Conversely, in orientation c the PAAEMA block is predominantly oriented as tails while the PMOEP block aligns, by electrostatic forces, along the solid substrate surface in a fashion well-known for the adsorption of oppositely charged polyelectrolytes. In neither orientation, however, would we expect the spectrum of the block copolymer adlayer to be identical to the spectrum of one or the other of the homopolymer adlayers, as in the ToF-SIMS vacuum environment the attached layers collapse and the impinging ions also probe trains, yet the signals from loops and tails are expected to be relatively more intense. In orientation a (Scheme 2) an even distribution of loops, trains, and tails in both of the blocks is envisaged, and one might expect a linear superposition of spectra. For the positive spectra, the loadings on PC1 contain no signals that can be attributed to the orientation of the block copolymer. This is due to the fact that the ions that dominate the loadings plot (Figure 4b) arise from both blocks. The loadings plots of Figures 6 and 7 show that the P-containing fragments contain significant information, which is not unexpected. These are key to the interpretation of the spectral differences between the three polymer adlayers. Quantitative interpretation of ToF-SIMS data is fraught with difficulty due to matrix effects, but with care and assuming that matrix effects are comparable, one can make quantitative comparisons for very similar surfaces. We shall assume that the two blocks emit ions in the same manner whether in the form of a homopolymer or the block copolymer; given the length of the blocks, this neglect of edge effects appears justified. On this basis, nonselective attachment (Scheme 2a) would be expected to lead to half the relative intensities of the P-containing fragment ions for the APS-PMOEP-b-PAAEMA copolymer sample compared with the APS-PMOEP adlayer, simply on the basis of halving the surface area occupied by the PMOEP block in the former. If the copolymer attaches onto the surface preferentially via the PMOEP block (Scheme 2c), we would expect the relative intensities of these fragments to be even less

1018 Langmuir, Vol. 25, No. 2, 2009

Figure 8. Statistically evaluated normalized intensities of the PO3- ion signal recorded on the PMOEP and PMOEP-b-PAAEMA polymer adlayer surfaces (average values across 10 samples, difference by Student’s t distribution R ) 0.05). The PO3- intensity is a measure of the surface density of the PMOEP segment, assuming similar ionization efficiencies in the two polymers.

than half compared with PMOEP. This arises because the PMOEP blocks would be partially covered by PAAEMA blocks and would thus be less exposed to the primary ion beam than the phosphates in the APS-PMOEP sample. If the copolymer attached via the PAAEMA block and the PMOEP block completely covered the PAAEMA block, we would expect POx ion intensities identical to those from the PMOEP adlayer. On the other hand, the positive ions (hydrocarbons and CxHyO) released from the two blocks appear to remain in the same relative ratios regardless of orientation as they are emitted from both blocks with apparently similar efficiencies. It may at first seem surprising that the relative intensities of the P-containing ions are, in fact, even greater for the APSPMOEP-b-PAAEMA sample than for the APS-PMOEP sample (Figure 7b). However, the spectra also contain signals from the substrate. Signal intensities vary due to the different amounts of attached polymer adlayers. The end result is that the effects from the different surface coverages are thus superposed on the chemical differences between the polymer adlayers. Unfortunately it was not possible to achieve identical surface coverages (expressed as identical substrate signal intensities in XPS) for the three polymer adlayers as their equilibrium adsorption amounts were different and attempts to force identical surface coverages would have come at the expense of equilibrium conformations. The XPS data (Table 1) suggest that the attached amount is somewhat higher for PMOEP-b-PAAEMA compared with the homopolymer adlayers. This needs to be factored into the consideration of relative ToF-SIMS ion intensities. Armed with the insight from PCA that the P-containing ions play the key role in spectral differences whereas the positive ions do not, we can now safely return to univariate analysis, plotting the relative (normalized) intensity of the PO3- ion signal (m/z ) 79) for APS-PMOEP and APS-PMOEP-b-PAAEMA in Figure 8. This shows that the relative intensity of that ion is ∼10% higher for the copolymer adlayer. Combined with the XPS information that the attached amount is ∼15% higher for the copolymer compared with the PMOEP adlayer, it would appear that the PMOEP-b-PAAEMA adlayer presents an ion yield not much less than that from the PMOEP adlayer. This suggests that the copolymer adlayer may have a definite preference for orientation b, with a substantial extent of segregation such that the PMOEP block is, on average, further from the solid surface than the PAAEMA block. The attenuation of the Si+ ion signal, on the other hand, is significantly greater than 15% (Figure 9). This reflects that, with the shallow ToF-SIMS probe depth,

Jasieniak et al.

Figure 9. Normalized ToF-SIMS intensities of the Si+ ion signal recorded on APS and polymer adlayer surfaces.

additional adlayer material rapidly attenuates substrate ion emissions. Thus, the contribution of ions from APS-PMOEPb-PAAEMA to the total integrated ion intensity is greater than for APS-PMOEP. Accordingly, the observed higher relative intensity of the PO3- ion for APS-PMOEP-b-PAAEMA is partly due to the fact that the copolymer:substrate ion intensity ratios are greater than the PMOEP:substrate intensity ratios. It appears reasonable to conclude that the contribution of substrate ions is insufficient to account for the observed higher relative PO3- ion intensity for the block copolymer. The copolymer contains only half the content of phosphate groups compared with PMOEP, and its adsorbed amount, as assessed from the XPS data, is significantly less than twice that of PMOEP; accordingly, to produce the observed phosphate ion intensities, there must be a degree of orientation of the PMOEP block away from the substrate surface. Hence, these analyses support the conclusion that conformation b best describes the adlayer but the extent of orientation appears to be only partial. Therefore, in summary, the ToF-SIMS analyses provide evidence for preferential attachment of the diblock copolymer via the keto groups of the PAAEMA block by reductive amination rather than via ionic bonding of the phosphate groups through a structure with some loops, trains, and tails rather than a totally flat conformation. The alternative possibility of attachment through phosphate groups binding to amine groups should have resulted in higher loadings of other peaks, particularly CxHyO+ ions assignable to the PAAEMA structure that would result preferentially in loops and tails. Taking into consideration drying effects in sample preparation and ion generation from within 1-2 nm probe depth, the exact conformation cannot be elucidated with greater precision. Nevertheless, it can be concluded that a conformation enriched in PMOEP tails applies. In addition, it appears that the outermost surface of the attached layer is richer in phosphate groups than the bulk composition. This in turn implies that under the conditions applied (dry DMF) the covalent coupling of keto groups to the amine groups of the aminated slide is more efficient than the proton transfer required for generation of an electrostatic attraction. This (partially) preferential orientation of the PMOEP-b-PAAEMA copolymer may have significant implications on interfacial interactions such as those implicated in the complex cascade of events that lead to nucleation and mineralization on such surfaces.

Conclusions The combination of PCA analysis and ToF-SIMS produced a powerful analytical technique for the study of surface-attached polymer layers and provided valuable and unique information on the conformation of attached block copolymer layers. This

Study of the Orientation of an Adsorbed Copolymer

study yielded novel information on the attached conformation of a PMOEP-b-PAAEMA copolymer, particularly with respect to the PMOEP segment orientation. Comparison with the pure homopolymer spectra and interpretation after application of PCA indicate that the adsorbed conformation is not random: the signals arising from the PMOEP block are relatively more intense than expected for a random conformation. This indicates an adsorbed conformation with a partial segregation toward PMOEP enrichment at the outermost surface. The fact that the polymers could not be coated at identical surface coverages complicated interpretation, and the first principal component contained a contribution arising from surface coverage. Nevertheless, taking into account the totality of spectra and analyses, we can conclude that the PMOEP-b-PAAEMA adlayer immobilized onto the APSmodified glass slide adopts a partially oriented conformation that tends toward conformation b shown in Scheme 2. The copolymer attaches to surface amine groups via the PAAEMA segment to a higher extent than via the PMOEP segment; thus,

Langmuir, Vol. 25, No. 2, 2009 1019

on average the PMOEP block is furthest from the APS surface. Finally this study demonstrates that ToF-SIMS is a powerful tool not only for investigating the surface composition of adsorbed layers, but also for probing the molecular conformations of adsorbed block copolymers. However, care must be exercised when PCA is performed on spectra acquired from multiple samples. As shown here the information in a three-way comparison of the three adlayers was somewhat ambiguous, whereas a pairwise comparison resulted in interpretable differences. Acknowledgment. Asper Biotech (AS) is acknowledged for supplying aminated slides. M.J. and H.J.G. acknowledge support by the Australian Government under the ARC Special Research Centres Scheme (Special Research Centre for Particle and Material Interfaces). M.M. acknowledges financial support from a QEII fellowship (ARC fellowship). LA802016B