10 min, to give the solution used for spin coating. All the chemicals were obtained from Aldrich Chemicals Ltd. Th e sols were dispensed on p-type, 75 mm diameter silicon wafers, through a 0.1 µm fi lter (PTFE Whatman, obtained from BDH Merk Ltd), and thereaft er the substrate was spun at 2000 rpm for 15 s. Th e coated substrate was baked at 100°C for 5 min, and then cleaved into 10 pieces. Each piece was baked in air at a diff erent temperature, in the range from 100 to 1000°C, for 30 min. Th e samples were kept in covered petri dishes for a few days in room conditions before the experiments were continued; this allows the completion of surface hydroxylation, and gave reproducible ellipsometer results when water is used as an adsorbate. Th e thickness and refractive index of the samples were measured using a Rudolph AutoEl III ellipsometer, with an
operating wavelength of 633 nm, and precisions of about ±0.002 and ±3 Å in index and thickness, respectively. For microporous fi lms, the measured index is strongly dependent on relative
humidity, because of condensation of water in the pores. By mea- suring the dependence of index on humidity, information about
porosity can be obtained. We have extended this technique to the use of diff erent adsorbate species, in order to probe pore sizes [3]; this, for the sake of brevity, we call molecular probe ellipsometry. In this technique, the fi lm is placed in a sealed chamber on the sample stage of the ellipsometer; fi rst dry N2 gas is passed through the chamber to empty the pores of any condensed adsorbate, and then N2 having been bubbled through the liquid adsorbate is passed over the sample to fi ll the pores; in each case the refractive index is measured. By assuming that all the accessible pores in
Methodology — Writing Task 69 dry and saturated atmospheres are completely empty or fi lled with adsorbate, respectively, the pore volume and index of the solid skeleton can be determined by an extension of the Lorentz-Lorenz relation [8] where nf, ns and np are the refractive indices of the fi lm, solid skeleton and pores, respectively, and vp is the volume fraction porosity. Measurement of nf for both the dry and satu- rated fi lms allows both vp and ns to be determined with the as- sumption that np has the same value as that of the bulk adsorbate in the saturated case, and of air (np = 1) in the dry case. In order to empty the pores, an initial high fl ow rate of N2 was used for a few minutes and the rate was then reduced to 1000 sccm (standard c.c per minute) for 15 min. the fl ow rate was kept at 100 sccm for 15 min to fi ll the pores. Th e low fl ow rate in this case reduces the likelihood of cooling of the sample surface, which could cause condensation on the external fi lm surface. Comparison of the measured fi lm thickness for wet and dry atmospheres indicated that this did not occur. Th e temperature inside the chamber was monitored by a thermocouple to ensure that there was no drift or alteration due to gas fl ow. In each case, the measurement was recorded once repeatable readings were obtained. Th e adsorbates used are listed in Table 2. Th eir average diameters were estimated using a combination of bond length data [9] and Van der Waals atomic radii [10]. All were obtained from Aldrich Chemical Ltd, except C24H44O8 obtained from Fluka Chemie AG. Th e optical quality of the fi lms was fi rst studied qualitatively by visual examination, and by optical microscopy. Th e homogeneity of the fi lms was then investigated quantitatively by measuring the intensity of scattered light resulting from oblique refl ection of a laser beam from the fi lm-coated silicon substrate. A helium-neon laser beam, having a wavelength of 633 nm, was directed onto the sample, through a chopping wheel, at an angle 59° from the normal. Th e specularly refl ected beam was absorbed onto a black card, and the scattered light was collected at normal incidence to the sample using a ×10 microscope objective, and measured using a silicon photodiode and a lock-in amplifi er. Th e position of lens and angle of incidence were fi xed during measurements.