Organosilanes are frequently applied as primers to oxide layers on aluminium and its alloys in order to improve adhesion to organic coatings, for example, for corrosion protection, although the bonding mechanism is not well understood [1, 2]. A common view has been that the adhesion involves formation of Si±O±Al type linkages, although direct evidence for this interfacial bonding has proved elusive [3±9]. Earlier work from this laboratory suggested that angle-dependent X-ray photoelectron spectroscopy (ADXPS), in combination with a bias-potential technique, may give a new way for probing the interfacial bonding between oxidized aluminium and silanes, and for aiding the identi®cation of relationships between atomic-level interfacial structure and macroscopic properties [10, 11]. For a thin layer of a-glycidoxypropyltrimethoxysilane (a-GPS) deposited on chemically etched aluminium, it was hypothesized that a new structure apparent in Al2p spectra for small exit angles in ADXPS, after applying a negative biasing potential, may indicate a ` chemical shift'' effect associated with the Si±O± Al bonding [10]. The evidence was indirect; it depended on the existence for some silane±aluminium systems of this additional structure in Al2p spectra, induced by differential charging, and an associated correlation with corrosion protection. This letter reports further observations on the same system, and includes an attempt at characterizing the interfacial bonding with secondary ion mass spectrometry (SIMS) in the static SIMS mode [12]. Following the earlier reports [10, 11], square panels of 7075-T6 aluminium (1 cm) were degreased and acid etched before coating with the silane solution (1 vol % formed by dissolving a-GPS monomer in an equi-volume mixture of distilled water and methanol). Three main samples are referred to below. Sample A was a blank sample prior to coating with aGPS (i.e. the alloy after etching); sample B was formed from sample A by applying the a-GPS solution drop-by-drop until a thick layer of polymer was formed; sample C gave the a-GPS=substrate interface formed by dipping sample A in the a-GPS solution for 5 min followed by air drying. X-ray photoelectron spectroscopy (XPS) spectra were measured with a Leybold MAX200 spectrometer [13] using the unmonochromatized AlKa source (1486.6 eV) operated at 15 kV, 20 mA with pressure in the analysis chamber around 6 3 10y7 Pa. The biased spectra were obtained by applying an external potential of y93 V through the sample holder, and then, after measurement, mathematically shifting the energy scale back by 93 eV [10]. Static SIMS was performed with a VG MM12-12S quadrupole mass spectrometer using a xenon primary beam (5 keV impact energy), for which a current of 0.2 nA irradiated a square area (6.25 mm) on the samples. XPS con®rms that the samples A, B and C had the expected compositions for the surface regions; also sample B was suf®ciently thick that Al signals were not detectable. Sample C showed the extra structure reported previously in the Al2p spectra from using the bias potential technique, and therefore it appeared to provide a suitable sample for probing the a-GPS=aluminium interface with SIMS. Relevant positive ion static SIMS spectra are reported in Fig. 1. Those for the acid-etched alloy (sample A) and the thick a-GPS layer (sample B) serve as references for interpreting that from the interface structure (sample C). The spectrum from sample A in Fig.1a is made up of Al (mass 27) and CH fragments, the latter being indicative of the commonly observed hydrocarbon contamination. Since C2H3 and C2H5 fragments are usually comparable in intensity, only part of the mass 27 peak can be assigned to C2H3 . The small peak at mass 28 may be AlH (Si is considered less likely since no Si was detected by XPS). Sample B experienced charging under ion bombardment, and the spectrum in Fig. 1b was obtained with operation of an electron ood gun (500 eV) to compensate. The signals obtained were fairly weak; nonetheless peaks from Si and SiOH, and possibly SiCH3 , were evident in the spectrum. The presence of SiOH is consistent with expectation from the a-GPS polymerization [3]. There should be no contribution at mass 27 from Al given the thickness of the a-GPS layer, but hydrocarbon fragments, in part from the a-GPS backbone, are clearly present. Fig. 1c shows the static SIMS spectrum for sample C. No sample charging was evident, thereby supporting the belief that the a-GPS layer was very thin. The most prominent peak was at mass 45 (SiOH from the a-GPS polymerization), but there