ANALYST, MAY 1993, VOL. 118 463 Surface Modification of the Biomedical Polymer Poly(ethy1ene terep h t h a late) Ldn N. Bui and Michael Thompson* Department of Chemistry, University of Toronto, 80 Saint George Street, Toronto, Ontario, Canada M5S ?A I Neil B. McKeown,+ Alex D. Romaschin and Peter G. Kalman Vascular Surgery Division, Toronto General Hospital, 200 Elizabeth Street, Toronto, Ontario, Canada M5G 2C4 X-ray photoelectron spectroscopy was used to characterize modified surfaces of a biomedically important polymer, poly(ethy1ene terephthalate). Several modification schemes were investigated and direct silanization with 3-aminopropyltriethoxysilane was found to be the optimum procedure, resulting in an aminated surface. Surface coverage of up to 100% was achieved with retention of the polymeric structural integrity.Further activation of the silanized surface was accomplished with two cross-linkers, glutaraldehyde and sebacoyl chloride. A simple biomolecule, L-cysteine, was successfully immobilized onto a surface pre-treated with 3-aminopropyltriethoxysilane and glutaraldehyde, with a coverage of 42%. Keywords: Biomaterial: biocompatibility; poly(eth ylene terephthalate); silanization; X-ray photoelectron spectroscopy An important aim in the field of biomaterial technology is the improvement of the biocompatibility of existing synthetic vascular grafts. Currently, the most common arterial substi- tutes are Dacron, composed of woven or knitted fibres of poly(ethy1ene terephthalate), (PET), and Gortex, produced from expanded poly(tetrafluoroethy1ene) (PTFE).These graft materials possess the required mechanical properties and excellent chemical stability in vivo. However, both types of graft are susceptible to occlusion as a result of thrombus formation,’ particularly under low flow, high resistance conditions present in small diameter arterial by-passes. Poly(ethy1ene terephthalate) is inherently more thrombogenic than PTFE;2?3 therefore, its primary application is restricted to larger diameter by-passes in the coronary area. The high flow rate and low resistance in this region tend to minimize thrombus formation. Poly(tetrafluoroethy1ene) is the conduit of choice for smaller diameter arterial by-passes. The cumula- tive five years patency for one such type of operation, a lower extremity by-pass, is only 17~25% .I The usual consequence of a graft failure is the amputation of the diseased appendage; therefore, substantial improvements in the biocompatibility of both graft materials are clearly required.Despite its greater thrombogenicity , PET is still favoured for large diameter arterial reconstructions because of its excellent handling characteristics during an operation. Its superior elasticity and pliancy make PET the preferred material, provided that its thrombogenicity can be improved to or beyond that of PTFE. It has been demonstrated that the surface properties of any graft material are of prime importance in the determination of its biocompatibility;4.5 hence a decrease in thrombogenicity may be achieved through modification of the surface of the polymer.There have been numerous attempts to attain this objective with the common approach being the immobiliza- tion of anti-thrombogenic moieties onto the surface of the polymer.6 Numerous compounds have been investigated, among them heparin,’ various prostaglandins8 and albumin.9 The nature of these attachments has largely been by physical adsorption. This has resulted in only limited success as the adsorbed compounds can be slowly removed by the blood flow. The stability of the modification will be enhanced if the inclusion of chemical covalency in all aspects of the modifica- * To whom correspondence should be addressed. + Present address: Department of Chemistry, University of Man- Chester, Manchester, UK. tion process can be effected.A different approach is the modification of the existing material so that it ‘mimics’ that of a biological membrane, a biomimetic surface.10 A natural cell membrane consists of a bilayer of phospholipids, in which proteins and glycoproteins are supported. Model films of phospholipids have been prepared using the Langmuir- Blodgett technique and such films have been shown to be generally thrombo-resistant .I1 These films are, however, too unstable to be transferred effectively onto any prostheses; hence more stable lipids and lipid-like surfaces must be developed. These can then be immobilized onto the polymer to provide non-thrombogenic surfaces. Other constituents of a membrane, such as those found in endothelial cells, including cholesterol and polymeric saccharides, can also be incorpor- ated.8 The inclusion of chemical covalency is still of prime importance because, unless a covalent attachment of the lipid layer can be achieved, any gain in biocompatibility would be short-lived.The lack of indigenous reactive functional groups on the PET surface precludes any covalent attachment unless the introduction of free, reactive chemical species onto the surface can be achieved. The functionalization of the polymer surface will allow further modifications of a more permanent, covalent nature. The target functional groups are hydroxyl or amino groups as both can facilitate further chemistry. Poly- (ethylene terephthalate) is a polyester; hence it is susceptible to a variety of chemical reactions at the carbonyl sites. Dave eta1.12 reported the saponification of PET, giving rise to a surface comprised of carboxylic and hydroxyl functional groups. Aminolysis using multifunctional amines to create free NH2 moieties was investigated by Avny and Reuben- feld.13 Multifunctional reagents were employed to reduce the reactivity and control the degree of polymeric degradation.The derivatization of PET with poly(ethy1ene glycol), using a diamine as a cross-linker, was reported by Kim and K0.14 Poly(ethy1ene terephthalate) can also be reduced using a strong reducing agent such as a metal hydride. The reduction of PET with LiAlH4, yielding a hydroxylated surface, was patented by Collins.15 In this investigation, all of these procedures were attempted and the resultant surfaces were analysed to determine which method provides the optimum modification. As these reactions involve the cleavage of the ester bonds, there exists a possibility of structural degrada- tion.This aspect was investigated for each experiment through the implementation of a degradation reaction, i.e., modifica-464 ANALYST, MAY 1993, VOL. 118 tion using an analogous but more concentrated solution or by simply adopting a longer reaction time. After functionalization has been achieved, the polymer can be further activated using various cross-linkers. This would facilitate the eventual immobilization of phospholipids or any other non-thrombogenic entities. An important class of coupling reagents are multifunctional silanes. The utility of these silanes can be attributed to the fact that they can form strong covalent bonds to a substrate and yet still retain their chemical reactivity.These silanes have been used to improve the bonding of polymer resins to metal surfaces and glass fibre. 16717 Several procedures have also been developed employing silanes and other cross-linkers to promote the covalent bonding of organic complexes to a functionalized surface. These include the development of a surface acoustic wave chemical sensor,18 the attachment of antibodies to a substrate for affinity chrornatography,'g the construction of a polymer-modified electrode20 and the immobilization of enzymes on solid supports for enzyme assay techniques .2l,Z2 Recently, Kallury et al.23 reported the covalent linkage of modified phospholipids to activated gold and glass surfaces, utilizing reactive silanes such as 3-aminopropyltriethoxysilane (APTES) and dichlorodimethylsilane (DCDMS) and also other cross-linkers such as sebacoyl chloride (SB) and monomethyl azelate.These procedures can be readily adap- ted for an activated PET surface as the initial active moieties would be the same, hydroxyl or amino groups. The immediate aim of this work was the adaptation of these methodologies to immobilize a simple biomolecule, L-cysteine, onto an acti- vated PET surface. X-ray photoelectron spectroscopy (XPS) was chosen as the primary means of analysis of the surfaces studied. Advancing water contact angle (WCA) measurement and scanning electron microscopy (SEM) were used to complement the XPS results. Experimental Materials and Chemicals The modifications were performed on thin films (0.25 mm) of PET (Mylar, DuPont, Toronto, Canada) (hereafter this will be referred to as surface 1) and on the manufactured woven Dacron graft (Meadox Medical, Oakland, NJ, USA) (surface 2).Owing to the prohibitive cost of the Dacron graft, most of the experiments were performed solely on the PET polymer. However, certain important experiments were performed concurrently on Dacron to ensure the applicability of the procedure to the actual graft material. All surfaces were cleaned by Soxhlet extraction with tetrahydrofuran (THF) for 4-6 h prior to and after all modifications to ensure the removal of most physically adsorbed contaminants. This was followed by rinsing with ethanol and chloroform. Toluene and THF were refluxed, prior to collection, over sodium and benzophe- none. Dichloromethane was distilled over phosphorus pento ide, also for several hours. All three solvents were distilled under a nitrogen atmosphere.Water was doubly distilled and de-ionized prior to use as a solvent and rinsing agent. Hydrazine, LiAlH4, HCI, 1,3-diaminopropane (DAPr), 1,5- diaminopentane (DAPe), tetraethylenepentaamine (TEPA), DCDMS, trifluoroacetic anhydride (TFAA), SB, 50% v/v glutaraldehyde (GA) in water, L-cysteine and all solvents were purchased from Aldrich (Milwaukee, WI, USA). Tetra- ethoxysilane (TES) and APTES were obtained from BDH (Toronto, Ontario, Canada). 3-Aminopropyldiethoxymethyl- silane ( APDEMS) and 3-aminopropylethoxydime thylsilane (APEDMS) were acquired from Fluka (Ronkonkoma, NY, USA).All silanes were vacuum-distilled prior to use and all reactions, except for base saponification, hydrolysis and cysteine immobilization, were carried out under a nitrogen atmosphere. Degradation experiments were performed with neat solutions of the specified reagents where possible. Modification Procedures Virgin PET (1) was incubated in a 10% v/v aqueous solution of hydrazine at 60 "C for 0.5 and 1.5 h to provide surfaces 3A and 3B, respectively. The degradation of each surface was achieved through an incubation period of 24 h in the same solution. Poly(ethy1ene terephthalate) samples (1) were incu- bated in 5% v/v toluene solutions of TEPA, DAPe and DAPr for 6 h under nitrogen. The results were the aminated surfaces 4,5 and 6, respectively. 1,3-Diaminopropane was also reacted with Dacron (2) for a period of 12 h to give surface 7.Degradation experiments were carried out using neat reagents. Virgin PET (1) was reduced using a solution of 0.5 g of LiAIH4 in 25 ml of dried THF for 20 min under nitrogen. The resulting sample, 8, was then acidified with 0.01 mol 1-1 HCI. The degradation experiment was carried out by incubating a sample in a similar solution for 1 h. All reactions with silanes were performed using 2% v/v toluene solutions under nitrogen atmospheres for periods of 24 h. The only exception was the silanization involving DCDMS, for which dichloro- methane was employed as the solvent. All silanized samples were cured for at least 48 h at ambient temperature and pressure prior to any subsequent modification.Samples previously reduced by LiAIH4 (8) were silanized with APTES and DCDMS to give surfaces 9 and 10, respectively. Both virgin PET (1) and virgin Dacron (2) were also silanized with APTES to give surfaces 11A and 12, respectively. An incubation time of 48 h was also used for PET to produce surface 11B. Silanized sample 11A was immersed in a buffer solution (pH 9) for 24 h, yielding surface 11C. Poly(ethy1ene terephthalate) (1) was also silanized with TES, APEDMS and APDEMS to give samples 13, 14 and 15, respectively. The degradation of PET was attempted by leaving samples in neat reagents for periods of up to 5 d. Poly(ethy1ene terephthalate) (1) and Dacron (2) were incubated in a solution of 2 ml of TES in 100 ml of dried toluene with 0.2 ml of DAPr added as an initiator.The reaction was carried out under a nitrogen atmosphere for a period of 24 h, resulting in surfaces 16 and 17. An unmodified PET sample (1) and silanized PET samples ( l l A , 14, 15 and 16) were suspended over neat TFAA in a vapour phase reaction for 24 h. The resulting surfaces were designated 18-22, respectively. A control PET (1), and an APTES-modified surface (11) were incubated in 5% v/v aqueous GA solution. The reaction was performed at ambient pressure and temperature for 24 h. The resulting surfaces were designated 23 and 24, respectively. A control PET sample (1) and an APTES-modified sample (11) were incubated in 3% v/v solutions of a second type of cross-linker, SB, to give specimens 25 and 26. Toluene was used as the solvent and 0.1 ml of pyridine was also added as a proton scavenger.The incubation time was 24 h and a nitrogen atmosphere was employed. Surfaces modified with APTES and APTES-GA (11 and 24) were incubated in a solution consisting of 1.0 g of L-cysteine dissolved in 10 ml of doubly distilled, de-ionized water for a period of 3 h. The resulting samples were designated 27 and 28, respectively. Instrumentation Some XPS spectra were acquired using an SSX-100 (Moun- tainview, CA, USA) electron spectroscopy for chemical analysis instrument at the University of Western Ontario (UWO) Surface Science Laboratory. The spectrometer was equipped with an unmonochromated aluminium K a source and the sampling area was 1 mm in diameter. A flood gun setting of 1.0 eV was required to compensate for charging effects.Scofield constants were used to derive sensitivity factors for the aluminium source [C(ls) = l.OO,O(ls) = 2.49, N(1s) = 1.68 and Si(2p) = 0.901.' Peak deconvolution wasANALYST, MAY 1993, VOL. 118 465 performed on a Hewlett-Packard (Palo Alto, CA, USA) 9836 computer using software provided by the manufacturer. All other spectra were taken using a Leybold MAX-200 surface analysis system (Leybold, Cologne, Germany) employing an unmonochromated magnesium Ka source with an excitation energy of 1253.6 eV. No flood gun was employed but the resulting spectra were calibrated by positioning the aliphatic carbon peak at 285.0eV. The observed charging effect is usually of the order of 3 4 eV. The sampling area utilized for all spectra was 4 x 7 mm.Survey and high- resolution spectra were obtained using pass energies of 192 and 48 eV, respectively. The elemental composition was calculated from satellite-subtracted spectra, normalized for constant transmission using sensitivity factors previously calculated for the Leybold MAX-200 system [F(ls) = 1.00, O(1s) = 0.78, C(1s) = 0.34, Si(2p) = 0.40 and N(1s) = 0.541. Peak deconvolution was performed using a program supplied by the manufacturer. Most spectra were obtained with the detector at 90" to the sample except during the angular dependency study, for which detector angles of 60,45 and 30" were employed. Consecutive area analyses of the carbon 1s peak of the control PET surface (surface 1) were performed to determine the extent of the damage to a sample caused by the X-ray source. The measured decrease in the intensity of the carbon peak was on average approximately 2% of the original peak, well within experimental error; hence, damage can be concluded to be minimal.The decrease may in fact be due only to the sputtering of adsorbed hydrocarbon contaminants by the X-ray source. Scanning electron microscopy was performed using a Hitachi (Tokyo, Japan) S-570 system with an accelerating voltage of 18 kV and a working distance of approximately 15 mm. The samples were gold-coated in a Polaron (Watford, Hertfordshire, UK) vapour deposition unit at 20 mA and 2.4 kV for a period of 120 s. Advancing WCAs were measured on a Remy-Hart (Mountain Lakes, NJ, USA) goniometer at 20°C. Measurements were taken for both sides of the advancing drop and the average was calculated.This was repeated three times for each sample to arrive at a more reliable mean. Results and Discussion Control Polymer The XPS survey spectrum of control PET (Fig. l), obtained using the XPS spectrometer at the UWO Surface Science Laboratory, yields the expected carbon (285 eV) and oxygen (532 eV) peaks (1 of Table 1). High-resolution analysis of the C(1s) peak reveals the three carbon bond types expected from an ester: carbonyl(289.0 eV), ether (286.5 eV) and hydrocar- bon (285.0 eV). The scanning electron micrograph shows a smooth, clean surface with only several white, circular artefacts [Fig. 2(a)]. These are either adsorbed dust particles or gold clusters, deposited during the coating process.The solid line near the top left-hand corner of the scanning electron micrograph is a photographic defect and not an inherent feature of the surface. The advancing WCA measure- ment of 82" (1 of Table 2) corresponds well with the reported literature range of 79-83.5".1*313 The survey and high-resolu- tion C( 1s) spectra of Dacron are almost identical with those of PET with the differences in percentage compositions being statistically insignificant (2 of Table 1). Scanning electron micrographs of the Dacron surface show the characteristic weaveknit pattern of the graft, the individual strands being smooth and even [Fig. 2(6) and ( c ) ] . The few white specks on the strands are again simply adsorbed dust particles or gold clusters. In view of the morphology of the surface, WCA measurements of Dacron could not be obtained for compari- son.The same PET and Dacron samples, analysed using the Leybold MAX-200 spectrometer, yield significantly different numerical values (1A and 2A of Table 1). This is probably 240 - (a) 220 - 200 180 - 160 - - - 60 - E 20 C 3 8 $ 280 *c 240 -r 260 2 220 5 200 - != 180 1 60 140 120 100 80 60 40 20 = 900 800 700 600 500 400 300 200 100 292 290 288 286 284 282 280 Binding energyleV Fig. 1 ( a ) Survey and (b) high-resolution C(1s) spectra of control PET Table 1 Elemental and high-resolution XPS of nucleophilic modifica- tions* Oxygen Nitrogen Surface Carbon (285.0 eV) (532.1 eV) (400.0 eV) 1. Control PET (UWO) Yo c 79.5 Yo 020.5 YO C-C 67.2 (285.0 eV) % C-0 17.2 (286.5 eV) YO C=O 15.8 (289.0 eV) 1A.Control PET (MAX-200) Yo C 71.2 Yo 0 28.8 Yo C-C 61.9 Yo C-0 20.7 Yo C=O 17.4 2. Control Dacron (UWO) Yo C 79.2 % 020.8 % c-c 67.4 Yo C-0 17.1 Yo C=O 15.6 2A. Control Dacron (MAX-200) %C 72.0 Yo 0 28.0 Yo C-C 62.1 Yo C-020.9 Yo C=O 17.0 3A. PET-hydrazine 3B. PET-hydrazine (0.5 h) YOC 75.3 Yo 023.4 Yo N 1.3 (1.5 h) % c 75.8 Yo 021.5 % N2.7 Yo c-c 59.9 YO C-0 22.8 Yo C=O 17.3 4. PET-TEPA Yo C 76.6 Yo 023.0 % N0.5 5. PET-DAPe YoC 75.1 Yo 023.9 Yo N 1.0 6. PET-DAPr % C 74.8 Yo 023.1 Yo N2.1 7. Dacron-DAPr % C 71.3 Yo 021.5 % N7.2 % C-C 65.0 Yo C-0 14.5 YO C=O 7.1 (288.9 eV) % C-N 5.9 (287.8 eV) * Binding energies for all high-resolution C(1s) data are similar to those of 1 except for figures accompanied by binding energies in parentheses.466 ANALYST, MAY 1993, VOL.118 Fig. 2 Scanning electron micrographs of control surfaces: (a) PET at magnification of X3000; (b) Dacron at a magnification of X 100; and (c) Dacron at a magnification of X3000 0 II Table 2 Advancing WCA measurements of selected surfaces Surface WCA 1. Control PET 82.0" I NH2 NHI 8. PET reduced using LiAIH4 63 .0" NH7 9. Reduced PET modified with APTES 67.5" NH2 NHz I NH2 NH2 NH2 I 10. Reduced PET modified with DCDMS 11. PET modified with APTES 13. PET modified with TES 14. PET modified with APEDMS 15. PET modified with APDEMS 76.8" 69.0" 77.0" 78.9" 71.2' I I 1 16. PET modified with TES/DAPr 27.0" NH 26. PET modified with APTES then SB 87.8" NH2 NH2 NH2 NH2 because different sensitivity factors were utilized for the elemental composition calculation. Of the two sets of values, the set obtained using the Leybold MAX-200 spectrometer (C = 71.2%, 0 = 28.8%) seems to be more consistent with the actual elemental composition of the monomeric structure (C = 71.4%, 0 = 28.6-/0).For meaningful comparison, each set of results will be referenced against its respective control data. Spectra were acquired at the UWO laboratory for surfaces 1, 2, 3A, 3B, 8, 9 and 10. Saponification With Hydrazine Saponification with hydrazine did not result in any significant degree of modification. A nitrogen peak at 400 eV is seen, but the amount detected is minimal, constituting only 1.3% of the over-all atomic composition (3A of Table 1). An increase in the reaction time produced a small increase in the elemental nitrogen, but again, not to any appreciable extent (3B of Table 1).A decrease of the carbon peak and a relative increase of the oxygen peak, compared with unmodified PET, can be explained by the fact that XPS is a surface-sensitive technique. Some of the carbon atoms on the surface may have been masked by the hydrazine, while the more mobile hydroxyl ends are closer to the surface and hence can be detected more easily. High-resolution analysis of the C( 1s) peak does not reveal an amide bond, but the -NCO peak may be too small and may be submerged beneath the carbonyl peak, which has a similar binding energy. Another possible explanation is that most of the observed hydrazine is not covalently bound but instead is physically adsorbed in the numerous pits and cracks of the modified polymer.Fig. 3 Mechanism of polymeric degradation It was observed that the original shiny, smooth PET was dulled and some scoring of the surface can be seen. This observation combined with the limited amount of modifica- tion led to the hypothesis that the observed phenomenon is the result of concurrent nucleophilic attacks at adjacent or neighbouring carbonyl sites. The subsequent multiple cleavages of the ester linkages can result in the loss of modified organic fragments (Fig. 3). This would lead to a peeling effect, i.e., any modified layer would be stripped off, leaving a fresh, unmodified polymer surface. This peeling phenomenon would effectively negate any modification generated by the saponifi- cation process. The degradation of the polymeric structure was confirmed by the observation that exposure of a PET sample to the hydrazine solution for 24 h resulted in complete destruction.The PET sample was reduced to tiny, barely visible fragments. It is clear that base saponification is not an applicable process of modification owing to this deleterious consequence. Aminolysis Avny and Reubenfeldl3 and Kim and K014 reported the introduction of free amino groups onto the surface of PET via reactions with organic diamines or higher amines. Aminolysis using 1 ,2-diaminoethane was also reported by Desai and Hubbe1124 as a method for introducing amino functional groups onto a PET surface, prior to the covalent binding of poly(ethy1ene oxide) (PEO). The aminolysis mechanism is simply a nucleophilic attack at the carbonyl site of the polymer by the amine. Modification of PET by this process was attempted, using DAPr, DAPe and TEPA, but the resultsANALYST, MAY 1993, VOL.118 467 were similar to those of hydrazine: minimal modification accompanied by extensive degradation of the polymer (4-6 of Table 1). All three aminated surfaces exhibit an increase in the percentage composition of carbon and a corresponding decrease of oxygen. This can be attributed to the masking effect of the carbon-containing amines. In the modification of PET, DAPr produced the largest nitrogen peak among the three amines, a still insignificant 2.1%. It is postulated that owing to its length, it is entropically favourable for DAPr to form a ring structure with the polymer through reactions at both of its NH2 ends.This would result in a greater amount of modification but the required primary amine functional group may now be in a less reactive secondary form. The failure of TEPA (0.5% N) seems to be in contradiction with results published by Avny and Reuben- feld. 13 It is possible that the nitrogen-containing entities detected by Avny and Reubenfeld were degraded PET fragments (Fig. 3) that were not properly rinsed off the surface, whereas our surfaces were more thoroughly extracted with THF after every modification. Extensive degradation of the sample was observed within 1 h, if neat solutions were employed, and within 48 h for 5% solutions. Scanning electron microscopy of the DAPr-modified PET surface reveals that the amine has degraded the polymeric structures considerably, stripping large fragments off the surface [Fig.4(a)]. Under higher magnifications, large craters within the polymeric structure can clearly be seen alongside the PET fragments [Fig. 4(b)]. The film became brittle and was easily torn. As it produces the most significant modification, DAPr was used to treat a Dacron surface. Initially, the result was encouraging as the amount of nitrogen observed increases to 7.2% (7 of Table 1). This indicates that a large number of amino groups had been introduced onto the surface. However, the scanning electron micrograph of this surface shows that the degree of degradation has also increased substantially [Fig. 4(c)]. A large number of the strands that make up the weaveknit pattern are damaged or completely severed, as a result of the modification.Under higher magnification, it can be seen that the polymer has 'melted' into a film, covering the surface [Fig. 4(d)]. This degree of degradation, undoubtedly, would greatly affect the tensile strength, and hence, the structural stability of the graft. The SEM results help to provide an explanation for the increase in the amount of amine. The Dacron graft is much thicker than the PET film; therefore, it can withstand more degradation, and hence, additional modification. The weaveknit structure of the graft also allows for the retention of some adsorbed amine, despite the intensive extraction process. It is clear that the increase in the amount of modification cannot compensate for the magnitude of degradation. Although the damage to the surface could probably be controlled by varying the experimental conditions such as concentration and incubation time, this method of modification is obviously unsuitable for the functionalization of the delicate Dacron graft material.Reduction using LiAlH4 This reaction should yield two alcohol functional groups per PET monomer. A significant decrease in hydrophobicity is observed for this surface (8), with the WCA measurement dropping from 82 to 63" (Table 2). This indicates the presence of polar groups on the surface. However, high-resolution C(1s) analysis discloses that the percentages of carbonyl and ether carbons of this surface (8 of Table 3) are not significantly different from those values obtained for the unmodified PET (1 of Table 1).In fact, the percentage of carbonyl carbons actually increases by 1.2%. A possible explanation is that the LiAlH4 may have cleaved the ester bond but did not completely reduce the resultant carboxylic acid. The alkoxide functionalities, created through ester cleavages, may have destroyed the metal hydride and interrupted the reduction process. The increase in the amount of oxygen and the corresponding decrease of carbon support the cleavage theory, as the oxygen-containing termini are more surface- oriented and hence can be detected more easily than the carbon backbone. Another possible explanation can be derived from the observation that the reduction process also affects the integrity of the polymeric structure. Extensive degradation of the sample was observed when the reduction time was extended for a few hours, indicating that the reduction may have only served to strip away the top layer of the PET surface, exposing a fresh, unmodified layer of PET.The resultant degradation, again, places limits on the utility of this procedure as a modification technique. Subsequent Reactions With DCDMS and APTES Reasonable amounts of APTES and DCDMS were found to have reacted with the reduced PET surfaces (9 and 10 of Table 3). The fact that more substantial silanizations are not observed can be linked to the limited success of the reduction process. The WCA measurements for the APTES- and DCDMS-modified samples are 67.5 and 76.8", respectively. These figures are higher than that of the reduced surface, and can be explained by the fact that the silanes on the surface are not as hydrophilic as the hydroxyl groups of the reduced sample.The APTES-modified sample contains polar amino groups, whereas the DCDMS-modified surface is comprised of highly non-polar methyl termini. Significantly more APTES is observed to be immobilized on the surface than DCDMS, in spite of the fact that DCDMS is considerably more reactive. It was also found that whereas a reasonable magnification of x 1000 amount of APTES and DCDMS reacts with the reduced PET, Fig. 4 Scanning electron micrographs of DAPr-modified surfaces: (a) PET at a magnifiLation of ~3000; (b) PET at a ma nification of x22 OOO; (c) Dacron at a magnification of ~ 2 2 2 ; and (&Dacron at a468 ANALYST, MAY 1993, VOL. 118 Table 3 XPS data for LiAlH4 reduction and subsequent silanization* 10.PET-LiAlH4-DCDMS 11A. PET-APTES (24 h) 11B. PET-APTES (48 h) 11C. PET-APTES (PH 9) 12. Dacron-APTES Oxygen Surface Carbon (285.0eV) (532.1 eV) 8. PET-LiAlH4 Yo C 75.2 Yo 0 23.8 YO C-C 64.3 % C-0 18.7 Yo C=O 17.0 Yo C-C 66.2 Yo C-0 20.7 Yo C=O 13.1 Yo C 68.8 Yo 0 26.2 Yo c-c 64.1 % C-0 19.5 Yo C=O 16.4 Yo C 65.8 Yo 0 22.3 Yo C-C 65.6 YO C-0 23.2 9. PET-LiAlH4-APTES % C 68.0 Yo 0 23.3 YO C=O 3.4 (289.3 eV) % C-N 7.7 (288.1 eV) Yo C 56.4 Yo 022.7 Yo C 68.5 Yo 023.2 Yo C 65.7 Yo 0 21.9 Yo C-C 69.0 Yo c-0 18.8 Yo C=O 6.9 YO C-N 5.3 (288.1 eV) % Si 5.1 Silicon Nitrogen (400.0 eV) (102.5 eV) YON 3.3 Yo N 0.7 Yo Si 4.3 % N 5.6 YO Si 6.3 YO NH2 44.8 (399.3 eV) % NH3+ 55.2 (401.1 eV) Yo N 10.2 Yo Si 10.8 Yo N 3.9 YO Si 4.5 YON 6.9 Yo SI 6.1 * Binding energies for all high-resolution C(1s) data are similar to those of Table 1 except for figures accompanied by binding energies in parentheses. an even larger amount of APTES is detected on the control, unmodified PET, silanized in the same reaction vessel (11A of Table 3).It is then apparent that more APTES is immobilized on the reduced surface because it can react with both the reduced and the unmodified areas on the surface, whereas DCDMS can only undergo a reaction with the hydroxylated portion. This unpredicted reactivity between APTES and virgin PET was investigated as it may provide a direct way to silanize a PET sample. Reaction With APTES The silanization of PET with APTES, without any prior reduction, results in the immobilization of a large amount of the silane (11A of Table 3).Surface coverage is calculated to be 79% for an incubation time of 24 h. A similar modification was obtained for the Dacron graft, with the amount of silane detected being of the same magnitude as that for PET (12 of Table 3). When the reaction time was increased to 48 h (11B of Table 3), surface coverage of 100% was achieved. Coverages of APTES on PET samples were calculated using observed intensities of the nitrogen peaks and the equation below:' where F is the percentage coverage, Ip is the intensity of the nitrogen peak, I, is the intensity of the nitrogen signal expected for a pure APTES layer of infinite thickness (7.1%), i . e . , the ratio of nitrogen atoms in an APTES molecule, d is the depth of the APTES layer, h is the escape depth of N(1s) electrons (25 A) and 8 is the angle of the detector with respect to the sample (90").As APTES tends to cross-link to form a multilayer siloxane network with a thickness well above the sampling depth of the XPS instrument (200 A), d can be set at It has been demonstrated that the deposition of APTES from an organic solvent usually results in the formation of a two- (monolayer) or three-dimensional (multilayer) poly- meric silane structure.25-27 Polymerization occurs via reac- tions between silyl ends of different APTES molecules to form 200 A. Si-0-Si bonds. Monolayer formation is caused by siloxane bond formation between surface-bound APTES molecules (horizontal polymerization). Multilayer generation is the result of reactions between surface-bound APTES molecules and solution-phase APTES molecules (vertical poly- merization). The thickness and stability of the silane poly- meric network depends greatly on the experimental con- ditions.The anhydrous, organic conditions employed in this work often result in thick multilayers with irregular mor- phology. It has also been determined that the curing process of an APTES-modified surface plays a very important role in the determination of its stability. Curing increases the number of siloxane bonds in the APTES film up to the maximum of three per monomer and can be carried out at ambient or elevated temperatures.28 For this investigation, curing was carried out under ambient conditions for at least 48 h prior to any subsequent modification.The curing process conferred a stability to the APTES layer immobilized on PET as Soxhlet extraction with hot THF and chemical rinsing with ethanol and chloroform failed to remove any significant amount of the silane. The modification is also fairly stable under aqueous conditions, although a loss of 30% of the silane is observed after immersion in a basic solution for 24 h (11C of Table 3). This figure is in agreement with previously published results.29 The fact that no further loss occurs suggests that only the loosely bound outer layers, which are either held together by weak forces (hydrogen bonds or hydrophobic interactions) or are incompletely polymerized (less than three siloxane bonds per monomer), are affected by an aqueous medium.The completely cross-linked inner layer is inert to nucleophilic attacks by the basic solution. In previous investigations, the mechanism for the attach- ment of the 'APTES film to the substrate has always been considered t o be a simple nucleophilic substitution, with the surface hydroxyl groups attacking the silyl group, displacing the ethoxide moiety and forming an 0-Si bond. This is also true for the reaction between APTES and the partially reduced PET. However, the reaction between APTES and the unreduced PET must proceed via a different pathway, as there is not a sufficient amount of nucleophile present on the unreduced PET, apart from a limited number of hydroxylANALYST, MAY 1993, VOL. 118 469 termini from the polymer chain ends. These termini are not present in any significant amount as indicated by the lack of reaction between unreduced PET and a highly reactive chlorosilane, DCDMS.It is proposed that the initial reaction occurs via nucleophilic attack by the amino group of the APTES molecule on the ester linkage of the PET in a manner analogous to the aminolysis reaction. The aminopropyl group of APTES can be considered to be comparable to that of DAPr. However, unlike all other aminolysis reactions, no degradation is observed even on prolonged exposure (5 d) to pure APTES. A secondary phenomenon must be responsible for the preservation of the PET polymeric structure, It is believed that the alkoxide group created by the ester cleavage may play a role in the preservation of the polymeric structure. A reaction between this anion and the silyl end of APTES, in a substitution reaction similar to the well-documented reaction between a hydroxyl group and APTES,18-23 would have the effect of holding both ends of the ester together. As the negatively charged alkoxide end of the cleaved ester is a much stronger nucleophile than a hydroxyl group, the second reaction is favourable.The proximity of the alkoxide moiety to the silyl group of the attacking APTES molecule favours a secondary reaction between these two moieties. This inserted APTES molecule would then cross-link horizontally and vertically with other APTES molecules near the surface to form the APTES flm. Not all silane molecules will be covalently attached to the PET surface but those that are act as anchors to bind the APTES overlayer to the substrate.This two-step mechanism is summarized in Fig. 5 . There are several other pathways by which the APTES coating can be anchored to the PET surface. The cleaved alkoxide end could react with a second APTES molecule, which may already be attached to the PET surface via its amine end, or is already part of the APTES polymeric network [Fig. 6(a)]. Another pathway may have the initial APTES molecule cross-linking with the APTES overlayer without undergoing the second step of the proposed two-step mechanism [Fig. 6(b)]. In either scenario, the integrity of the PET molecule would still be preserved as both ends of the cleaved ester are now bound to the APTES film. Two other pathways, not related to the proposed mechanism, may also exist.The APTES layer may be bound to the substrate via siloxane bonds with any existing hydroxyl groups on the surface, i.e., polymeric chain ends [Fig. 6(c)] or through hydrogen bonds between the amino groups and the ester EtO-S;, EtO' '*OEt I 0 II C- / 0 4 *+ EtO OEt oxygen [Fig. 6 ( d ) ] . All of the above schemes were examined in greater detail through reactions with several close derivatives of APTES and the results are discussed in a later section. High-resolution analysis of the C( 1s) peak for surface 11A by XPS (Fig. 7) reveals the presence of a new carbon bond type with a binding energy between those of the carbonyl and ether bonds (288.1 eV). This peak can be assigned to the carbons attached to the nitrogen atoms within the APTES layer (C-N).A marked decrease in the amount of carbonyl carbon (C=O) and a corresponding increase in the amount of ether carbon (C-0) are also observed. Although the mor- phology of the APTES layer is uneven, ranging from several to hundreds of monolayers thick,*5-27 its average thickness is significantly greater than that of the escape depth of electrons APTES multilayer 111111111111u111u1111*111111111111111111111111111111*1 (a) I Si I A PET substrate PET substrate APTES multilayer NH I c=o PET substrate APTES multilayer IIIUIIIIIIIIIIII1I111111111111111111111111IB111IIIIIIBIIIIWI H HN\ I 0 g PET substrate Fig. 6 Schemes of interaction between the APTES overlayer and PET; (a) linkage via the alkoxy group; (b) linkage via the amino group; (c) linkage via surface hydroxyl group; and ( d ) linkage via surface hydroxyl group; and ( d ) linkage via hydrogen bonding 200 180 160 140 120 100 80 60 1, 40 2? 20 8 1000 900 800 700 600 500 400 300 200 100 r. 3 N 0 7 200 > 'u, 180 $ 160 5 140 4- 4- 120 100 80 60 40 20 294 292 290 288 286 284 282 280 Bind i ng ene rgyIeV (a) Survey and (6) high-resolution C(1s) spectra of APTES- Fig.7 modified PET Fig. 5 Mechanism of APTES insertion into PET470 ANALYST, MAY 1993, VOL. 118 ~~~ ~~ Table 4 Angle-resolved XPS data of APTES-modified PET sample Detector angle c (Yo) 0 (Yo) N (Yo) Si (YO) 90" 65.8 22.3 5.6 6.3 60" 67.8 21.6 4.8 5.8 45" 66.6 20.8 6.0 6.8 30" 68.0 19.9 4.7 7.3 Fig. 8 Scanning electron micrographs of APTES-modified surfaces: (a) PET at a magnification of ~ 7 0 0 ; (b) PET at a magnification of x 14 800; (c) Dacron (2Y0, 24 h) at a magnification of X 10 0oO; and (d) Dacron (5%, 48 h) at a magnification of ~ 5 0 0 0 originating from the PET bulk.The decrease in the amount of carbonyl carbons can, therefore, be attributed to the shielding effect of the APTES layer. In fact, as the APTES layer does not contain any carbonyl bonds, the peak at 289.3 eV can be attributed to amide carbons, which have a similar binding energy. This statement is correct only if the coverage of the multilayer APTES formation is 100% , i.e., there is no exposed PET. Previous mathematical calculation places the coverage at approximately 79% ; therefore, some contribution from bulk PET is probable, with the balance being contributed by the amide bond at the PET/AFTES interface.The presence of the APTES layer also accounts for the increase in the ether carbon percentage as the APTES layer contains numerous ether-like carbons (C-0-Si). As the amino groups of the APTES layer are not involved in the cross-linking process, they are available for any subse- quent reaction. However, a certain proportion will be in the unreactive, protonated or H-bonded form.29 Protonation occurs via proton transfer from uncondensed silanols or ambient moisture while hydrogen bonding may exist between the amino groups and such silanols. High-resolution analysis of the nitrogen peak reveals that there are indeed two types of nitrogen species present on the surface, with binding energies of 399.3 and 401.1 eV, respectively. The peak at 399.3 eV can be attributed to free amine nitrogens while the 401.1 eV peak belongs to protonated amine nitrogens.As only the unproto- nated amines are able to participate in any future nucleophilic attack, it is clear that less than half of the amines present are in an active form. Pyridine was added in an attempt to scavenge free protons, but the concentration was kept low to prevent degradation of the polymer. An assessment of the over-all reactivity of the silanized surface through a reaction with TFAA was conducted prior to any subsequent immobiliza- tion. An angular dependence study was executed to determine the profile of the immobilized APTES layer but the results (Table 4) reveal that no angular dependency exists. The scanning electron micrographs are consistent with these results in showing the uneven topography of the siloxane polymer on the PET [Fig.S(a)]. Prominent on the APTES- treated surfaces are hemispherical 'islands' with diameters of less than 1 ym. These hemispheres are present singly or in linked aggregates. On closer inspection, it can be seen that the surface seems to be covered with a thin film of silane, out of which grow the macroscopic islands [Fig. 8(b)]. The SEM results are in agreement with a previous study of the polymerization structure of APTES .29 The origin of these macroscopic structures has not been conclusively determined but it is believed that they form in the reaction solution, prior to attachment to the APTES layer.29 Scanning electron micrographs of the Dacron surface treated with APTES show a similar pattern but the silane layer also exhibits a 'scab-like' appearance in addition to the hemispherical clumps [Fig.8(c)]. An interesting phenomenon is observed on a Dacron surface modified using a higher silane concentration and a longer reaction time. A coating of material can be seen covering most of the surface. This layer seems to serve as the anchor for a complex network of linked spheres [Fig. S(d)]. It is hypothesized that the outer network is the loosely bound portion of the APTES multilayer, some of which can be displaced by a basic aqueous medium. The basal portion is more strongly bound, rendering it stable in most media. Reactions With Other Silanes In an effort to elucidate further the mechanism of the PET-APTES interaction, the reactions between PET and several silane derivatives, structurally related to APTES, were examined.The first compound examined was TES, a silane in which the aminopropyl moiety has been replaced by another ethoxy group. The failure of the reaction between TES and PET (13 of Table 5 ) strongly suggests that the amino moiety does indeed participate in the reaction with PET. The minimal amount of TES detected can be attributed to the small amount of hydroxyl chain ends present on the surface of the polymer. The lack of reactivity suggests that the reaction between the silyl end of the silane and the hydroxyl termini does not play an important role in the anchoring of the APTES layer to the substrate [Fig. 6(c)]. Physical adsorption of the silane onto the substrate is also a possibility.This minimal amount of TES reduces the WCA measurement to 77" (Table 2). As TES polymerizes to a greater extent than APTES, the lack of reactivity between TES and PET also suggests that the initial attack by the amino group is the determining factor in the immobilization process. The adsorptiodpolymerization effect exerted by the silyl end of the molecule is a secondary process and by itself cannot account for the reaction between APTES and PET. The next silane examined was APEDMS. This molecule is similar to APTES, except that two of the three ethoxy groups have been replaced by inert methyl groups. As it has only oneANALYST, MAY 1993, VOL. 118 47 1 ethoxy entity that can be displaced, extensive polymerization is not possible with this compound. There are two possible routes for the reaction to proceed following the initial attack by the amino group: dimerization with another silane mol- ecule, which would result in degradation of the surface, or displacement of the ethoxy group by the alkoxy anion, which would preserve the polymeric structure. No degradation of the PET structure is observed, which eliminates the dimerization pathway and lends credence.to the hypothesized two-step mechanism. X-ray photoelectron spectroscopy data (14 of Table 5) reveal that only a small amount of APEDMS is actually bound to the surface. A reduction in WCA measure- ment to 78.9" seems to corroborate this observation. The reduced reactivity of the silane can be attributed to the steric hindrance caused by the substitution of the two ethoxy moieties with inert methyl groups.The lack of extensive polymerization also contributes to this reduction as the degree of immobilization is now primarily dependent on the reactivity of the amino group. The result shows that the amino functionality is reactive only to a moderate degree. This can be attributed to the fact that APTES is known to form an internal zwitterion, through the co-ordination of the amino group with the silyl skeleton, in the presence of trace amounts of water.30 The nucleophilicity of the amino group would be diminished as the lone pair is involved in the co-ordination. High- resolution analysis of the C(1s) peak shows slight decreases in the amount of ether and carbonyVamide carbons. This corresponds to a thin, incomplete coverage of the silane.As the third APTES derivative, APDEMS, has one less ethoxy group than APTES, the degree of polymerization should be reduced; hence a thinner and more even coverage can be achieved. This is indeed the observed result as demonstrated by XPS data (15 of Table 5). High-resolution analysis of the C(1s) peak reveals a trend similar to that of APTES, but with a smaller magnitude. This agrees with the result of the survey spectrum and confirms the presence of a thinner layer of silane. The reduced polymerization is not detrimental to any great extent as the top layer is loosely bound and can be easily solubilized chemically. The WCA measurement of 71.2" is close to that of the APTES-modified sample; hence it may be postulated that the number of NH2 groups present at the surface may be close to that for APTES.The success of this reaction provides a cleaner method of silanizing a PET surface. However, as high-resolution analysis of the nitrogen peak shows that only 34.6% of the amines are in the active unprotonated form, this surface may not be as reactive as that modified with APTES. The reactivity of this silanized surface was determined through a subsequent reaction with TFAA and the results are discussed in a later section. Reaction With TES, Initiated by DAPr As the APTES molecule may be thought to possess two distinct reaction sites, the aminopropyl end and the silylethoxy end, the combination of a molecule containing an amino- propyl group, DAPr, and TES should provide a reaction medium analogous to that of APTES, with the important difference being that the two functional groups are not connected. Such a reaction was attempted and the XPS results (Fig.9 and 16 of Table 5) reveal that a considerable amount of TES is immobilized on the surface of the polymer. This layer of TES was hydrolysed by atmospheric water vapour to yield a surface comprised of silanol groups. The presence of these silanol functions is substantiated by an extremely low WCA measurement of 27". Another observation acquired from this experiment is the fact that the surface of the PET shows obvious signs of degradation. Recalling that TES does not react with unmodified PET in the absence of an initiator, it seems that the immobilization of TES occurs only because the initial aminolysis by the diamine creates a reactive alkoxy anion, which can then displace the ethoxy group on the TES molecule. This initiation process appears to play a critical role in the success of the reaction, even though only a minimal amount of the amine was required.The degradation of the PET can be explained by the fact that the two reactive components are not connected together as with APTES. The integrity of the polymeric structure is compromised because, of the three pathways that could have preserved the polymeric structure, two have been eliminated [Figs. 5 and 6(b)], leaving only one [Fig. 6(a)]. The degree of degradation is minimal, which may be attributed to the action of the third pathway and the low concentration of the amine. High-resolution analysis of the C(1s) peak shows a trend similar to that of APTES, a reduction in carbonyl carbons and an increase of ether carbons, consistent with the presence of the silane.However, no C-N peak is detected as the amount of amine is minimal and any initial amide formation would be rapidly masked by the layer of TES. The large amount of silane detected suggests the presence of a thick multilayer, consistent with the greater polymerization capability of TES. Nevertheless, the fact that silanol groups are present on the surface has important ramifications as these provide an alternative route for further chemistry. However, Table 5 XPS data for silanizations involving APTES derivatives* Surface Carbon (285 .O eV) 13. PET-TES Yo C 71.2 14. PET-APEDMS Yo C 73.0 Yo C-C 67.6 Yo c-0 18.1 Yo C=O 14.3 15.PET-APDEMS Yo C 70.2 Yo C-C 70.2 % c-0 18.1 % C=O 8.3 % C-N 3.4 16. PET-TES/DAPr Yo C 36.9 Yo C-C 64.6 Yo C-0 24.1 Yo C=O 11.3 17. Dacron-TES/DAPr % C 26.4 % C-C 68.2 % C-0 26.6 Yo C=O 5.2 Oxygen (532.1 eV) Nitrogen (400.0 eV) Silicon (102.5 eV) Yo 0 26.6 Yo N 5.6 'YO Si 0.5 Yo 024.2 YON 1.4 YO Si 1.4 Yo 0 18.2 YON 5.1 % Si 6.2 YoNHZ 34.6 YoNH,+65.4 Yo 045.7 %N 2.5 YO Si 14.9 % 049.0 % N 2.2 % Si22.4 * Binding energies for all high-resolution C(1s) data are similar to those of Table 1.472 ANALYST, MAY 1993, VOL. 118 tertiary silanols may not be very reactive, owing to steric constraints; hence a reaction with TFAA is required to determine the extent of their reactivity. A similar modification was carried out on a Dacron surface. The result is even more significant as the atomic composition of Si reaches the extremely high figure of 22.4% (17 of Table 5 ) .This reflects a high degree of polymerization, which may or may not be suitable for a modification process, but it can be controlled through variation of experimental condi- tions. High-resolution analysis of the C( 1s) peak displays the same trends as those of the PET modification, but to a greater degree. Assessment of Reactivity Using TFAA A subsequent reaction between the silanized surfaces and TFAA was performed to assess the reactivity of the amino nucleophiles on these surfaces. A significant amount of the trifluoroacetate moiety is detected on the surface pre-treated with APTES (19 of Table 6), which attests to the high chemical reactivity of the NH2 groups.The apparently anomalous increases in elemental nitrogen and silicon are due to the use of a different pre-modified sample. High-resolution analysis of the C(1s) peak shows a large, distinct C-F peak at 500 450 400 350 300 250 200 150 2 50 8 900 800 700 600 500 400 300 200 100 1, 100 C 3 900 800 700 600 500 400 300 200 100 Binding energyleV Fig. 9 uninitiated and (b7 initiated with DAPr Survey s ectra of PET surfaces modified with TES, (u) 293.1 eV. The carbonyl peak also increases in size, which corresponds well with the presence of the trifluoracetate groups on the surface. X-ray photoelectron spectroscopy data for the reaction between the APEDMS-modified PET sample and TFAA reveal that TFAA is immobilized on this surface (20 of Table 6) to the same minimal degree as that of the control (18 of Table 6).This indicates that the surface-bound amino groups may be in an inactive conformation. One such conformation is the amide bond, formed in the initial reaction. A subsequent dimerization through the silyl end would have regenerated the reactive amine group. The lack of reactivity seems to exclude the dimerization pathway and supports the hypothesis that the ethoxy group present may have reacted with the cleaved alkoxide end of the PET monomer. Analysis by XPS of the APDEMS-modified surface reacted with TFAA (21 of Table 6) confirms that the surface-bound amino groups are chemically reactive although to a lesser degree than for the APTES-treated surface. This can be explained by the smaller number of amino groups and the higher proportion of the unreactive protonated form present on the former surface (15 of Table 5).Virtually no trifluoroacetate is detected on the surface modified with TES and TFAA (22 of Table 6), suggesting the failure of the tertiary silanols to act as nucleophiles. This supports the contention that TFAA is covalently bound to the aminated surfaces as opposed to being physically adsorbed onto the polymeric silane network. If physical adsorption is the mechanism of attachment, a significant amount of TFAA would have adsorbed onto the extensive siloxane network on the TES/DAPr sample. Reactions With Cross-linkers After the reactivity of the APTES-modified surfaces had been verified, a common cross-linker, GA, was reacted with this surface in an effort to facilitate further linkages.This di-aldehyde molecule, often used for protein coupling, reacts directly with the amino groups on the silanized surface to form a Schiff's base. The other terminal aldehyde can form a second Schiff's base with the amino moiety on any amino acid or protein. An APTES-modified sample, subsequently treated with GA, shows decreases in the nitrogen and silicon signals and an increase in the oxygen composition (24 of Table 7). This corresponds to the immobilization of a layer of the GA on top of the silane surface. More importantly, the control PET surface shows no such deposition, substantiating the covalent nature of the reaction. Surface coverage cannot be determined owing to the lack of a label element in the GA molecule but a sub-monolayer coverage is a probable result as it has been determined that only half of the amino groups in the APTES Table 6 XPS data for reactivity assessments with TFAA Carbon Oxygen Nitrogen Silicon Fluorine Surface (285.0eV) (532.1 eV) (400.0eV) (102.5eV) (688.4eV) 18.PET-TFAA Yo C 69.9 Yo 028.3 - 19. PET-APTES-TFAA % c 40.7 yo 020.5 % N6.5 % Si 8.8 % F23.4 - Yo F 1.6 Yo c-c 44.6 Yo C-0 17.0 Yo C=O 16.0 Yo C-N 6.7 % C-F 14.1 (293.1 eV) 20. PET-APEDMS-TFAA Yo C 70.8 % 024.5 21. PET-APDEMS-TFAA Yo C 61.9 Yo 023.0 YO C-C 61.1 Yo C-0 19.2 Yo C=O 16.5 % C-F 3.2 22. PET-TES/DAPr-TFAA YO C 31.0 YO 0 47.4 Yo N 2.3 YO Si 19.3 Yo F 0.0 % N 1.4 %Si 1.1 Yo F 1.3 Yo N 2.7 YO Si 2.4 Yo F 7.4ANALYST, MAY 1993, VOL. 118 473 Table 7 XPS data for reaction with cross-linkers and L-cysteine* Carbon Oxygen Nitrogen Silicon Chlorine Sulfur Surface (285.0 eV) (532.1 eV) (400.0 eV) (102.5 eV) (199.6 eV) (163.6 eV) 23.PET-GA Yo C 71.0 % 029.0 - - - - 24. PET-APTES-GA Yo C 65.7 % 024.6 % N4.5 Yo Si 5. I - - YO C-C 68.8 Yo C-0 19.2 YO C=O 10.9 (288.0 eV) YO C-N 3.7 (290.9 eV) 25. PET-SB Yo C 71.8 Yo 0 28.2 26. PET-APTES-SB Yo C 58.8 Yo 0 17.4 Yo C-C 72.3 Yo c-0 18.0 Yo c = o 5.5 YO C-F 4.2 Yo c-c 60.8 Yo C-0 26.0 Yo C=O 13.2 27. PET-APTES-cysteine YO C 63.9 Yo 0 28.8 28. PET-APTES-GA-cysteine YO C 60.1 Yo 022.1 Yo N 8.9 YO Si 2.0 - Yo s 7.0 Yo C-C 53.2 Yo C-0 29.6 Yo C=O 17.3 (288.3 eV) - - % c10.0 Yo N6.4 % Si9.9 Yo CI 7.5 - YO Si 1.8 Yo N4.1 % S1.4 * Binding energies for all high-resolution C(1s) data are similar to those of Table 1 except for those followed by figures in parentheses.200 150 7 100 v) c 3 0 v) 50 4 4 900 800 700 600 500 400 300 200 100 m 2 320 .z 280 2 240 200 160 120 80 40 . > S C - 292 290 288 286 284 282 280 Binding energylev Fig. 10 (a) Survey and (6) high-resolution C( 1s) spectra of immobi- lized 1.-cysteine on an APTES-modified PET surface, cross-linked with GA layer are in an active conformation. High-resolution analysis of the C(1s) peak reveals a reduction in the ether peak and a large increase in the carbonyl signal, again consistent with the presence of the aldehyde (24 of Table 7). The carbonyl signal was shifted to a lower binding energy of 288.0 eV, characteris- tic of an aldehyde functionality. A small peak is seen at 290.9 eV, which can be assigned to the imine linkage of the Schiff's base.This signal is smaller than that of the free aldehyde owing to the masking effect of the GA layer. A second type of cross-linker, SB, was also employed in the conversion of the amino groups of the APTES surface into more reactive sites. The amino group can form an amide bond with one end of the di-acid chloride, leaving the distal acid chloride free for subsequent nucleophilic attack. The success of this experiment is confirmed by the XPS results. Survey data (26 of Table 7) establish the presence of the acid chloride on the APTES-treated surface and confirm the inertness of the same molecule to the control PET sample (25 of Table 7). High-resolution analysis shows an increase in the carbonyl peak, consistent with the presence of the distal acid chloride moiety and the amide group.Reductions in the C-0 and C-N contributions also support this observation. A WCA measure- ment of 87.8" reflects an increase in hydrophobicity caused by the introduction of the non-polar Cx chains onto the surface. Reactions With L-Cysteine A PET surface, initially modified with APTES, does not couple to L-cysteine to any significant extent (27 of Table 7). The low reactivity between the amino group on the silane and the carboxylic function of the amino acid clearly demonstrates the need for a more reactive cross-linker. A successful coupling is achieved with a similar surface, after treatment with GA. A considerable amount of L-cysteine is bound to the surface as demonstrated by a 7% sulfur peak (28 of Table 7).An approximation, using eqn. (1), places the surface coverage of the amino acid at 42%. An increase in the nitrogen signal and a decrease in the Si peak substantiate the presence of a surface-bound cysteine layer. High-resolutim analysis o f the C(1s) peak reveals a further increase in the carbonyl bond type (Fig. lo), compared with surface 24, which agrees well with the introduction of the carboxylic acid-containing L-cysteine. The calculated cysteine coverage Of 42% indicates that the loss of any previous modification is minimal (assuming an initial APTES coverage of loo%, 45% of which are in an active conformation, and a 1 : 1 : 1 reaction ratio of APTES, GA and L-cysteine). This suggests that the APTES + G A surface is sufficiently stable in an aqueous medium to ensure the success of any further immobilization.This inertness to an aqueous environment is a requirement for any modification to a biological implant . The immobilization of I>-cysteine was not attempted on the surface pre-modified with APTES and SB, as an aqueous reaction medium, necessary for the solvation of the highly polar cysteine molecule, would result in the hydrolysis of the acid chloride. The SB surface can be employed for the immobilization of molecules that are soluble in a non-polar solvent, such as various phospholipids, as outlined by Kallury et a1.23474 ANALYST, MAY 1993, VOL. 118 Conclusions The stated aim of creating reactive surfaces on which the covalent immobilization of possible non-thrombogenic enti- ties can take place has been achieved.Aminated surfaces were generated via direct silanization with APTES and APDEMS, with surface coverages of up to 100%. A surface consisting of silanols has also been created via reaction with TES, initiated by DAPr. These modifications have been shown to be stable in most chemical environments; hence application to clinical uses is feasible. Reaction mechanisms between PET and the silanes have also been postulated. It is hypothesized that the silane films are anchored to the PET surface via several interactions. Pathways that involve a nucleophilic amino functionality play a more important role than those that make use of just the siloxy end of the molecule. Glutaraldehyde and SB were employed to activate the aminated surface further.A simple biomolecule, L-cysteine, was successfully immobilized onto a surface pre-treated with APTES and GA. The surface coverage of L-cysteine was calculated to be 42%. The surface modified with SB is unstable in an aqueous medium owing to the high reactivity of the acid chloride moiety to water; hence this surface is reserved for future immobilization of non-thrombogenic species that are soluble in a non-polar solvent. This work was supported by the Medical Research Council of Canada and the Physicians’ Services Incorporated Founda- tion.The authors also thank Drs. K. M. R. Kallury, D. C . Stone, S. Vigmond and M. Yang of the University of Toronto for helpful discussions during the preparation of this manu- script. Thanks are also due to Dr. R. Sodhi for assistance with the XPS experiments. References Anderson, J. M., and Kottre-Marchant, K., CRC Crit. Rev. Biocompatibility, 1988, 1, 111. Lord, R. S. A., Nash, P. A., Raj, B. T., Stary, D. L., Graham, A. R., Hill, D. A., Tracy, G. D., and Goh, K. H., Ann. Vascular Surg., 1988, 2, 248. Kottre-Marchant, K., Anderson, J. M., Miller, K. M., Mar- chant, R. E., and Lazarus, H., J. Biomed. Muter. Res., 1987, 21, 379. Andrade, J. D., in Surfaces and Interfacial Aspects of Bio- medical Polymers, ed. Andrade, J. D., Plenum, New York, Ratner, B. D., Johnston, A. B., and Lenk, T. J., J. Biomed. Muter. Res., 1987, 21, 59. 1985, ch. 5, pp. 105-191. 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Jozefowicz, M., and Jozefowicz, J., Pure Appl. Chem., 1984, 56, 1335. Barbucci, R., Castellani, C. B., Delfini, M., and Ferruti, P., Inorg. Chim. Acta, 1984, 93,47. Llonos, G., and Sefton, M. V., Biomaterials, 1988, 9, 429. Kottre-Marchant, K., Anderson, J. M., Umemura, Y., and Marchant, R. E., Biomaterials, 1989, 10, 147. Durrani, A. A., and Chapman, D., in Polymer Surfaces and Interfaces, eds. Feast, W. J., and Muno, H. S., Wiley, New York, 1987, p. 189. Hayward, J. A,, and Chapman, D., Biomaterials, 1984, 5 , 135. Dave, J., Kumar, R., and Srivastava, H. C., J. Appl. Polym. Sci., 1987, 33, 155. Avny, Y., and Reubenfeld, L., J. Appl. Polym. Sci., 1986,32, 4009. Kim, K., and Kop, S., J. Appl. Polym. Sci., 1986, 32,6019. Collins, R. J., US Pat. 2 955 954, 1964. Schmidt, R. G., and Bell, J. P., in Advances in Polymer Science, ed. DuSek, K., Springer-Verlag, Berlin, 1986, vol. 75, pp. 33- 71. Rosen, M. J., J. Coatings Technol., 1978, 50, 70. Heckl, W. M., Marassi, F. M., Kallury, K. M. R., Stone, D. C., and Thompson, M., Anal. Chem., 1990, 62, 32. Sportsman, J. R., and Wilson, G. S.. Anal. Chem., 1980, 52, 2013. Kannuck, R. M., Bellam, J. M., and Durst, R. A., Anal. Chem., 1988,60, 142. Loscascio-Brown, L., Plant, A. L., and Durst, R. A., Anal. Chim. Acta, 1990, 228, 102. Schifreen, R. S., Hanna, D. A., Boners, L. D., and Carr, P. W., Anal. Chem., 1977,49, 1929. Kallury, K. M. R., Ghaemmaghami, V., Krull, U., Davies, M. C., andThompson, M., Anal. Chim. Acta, 1989, 225, 369. Desai, N. P., and Hubbell, J. A., J. Biomater. Res., 1991, 25, 829. Kallury, K. M. R., Krull, U. J., and Thompson, M., Anal. Chem., 1988,60, 169. Pluddeman, E. D., in Silylated Surfaces, eds. Leyden, D. E., and Collins, W. T., Gordon & Breach, New York, 1980, p. 31. Murray, R. W., in Silylated Surfaces, eds. Leyden, D. E., and Collins, W. T., Gordon & Breach, New York, 1980, p. 125. Boerio, F. J., Schoelein, L. H., and Greivenkamp, J. E., J. Appl. Polym. Sci., 1978. 22, 203. Vandenberg, E. T., Bertilsson, L., Liedberg, B., Uvdal, K., Erlandsson, R., Elwing, H., and Lundstrom, I., J. Colloid Interface Sci. , 1991, 149, 1. Chiang, C. H., Lui, N. I., and Koenig, J. L., J. Colloid Interface Sci., 1982, 86, 26. Paper 2106339E Received November 26, 1992 Accepted January 13, 1993