Scratching the surface with neutrons - reflectometry at HIFAR

Andrew Nelson, Michael James and Jamie Schulz

The new HIFAR Reflectometer is now in routine operation and is used for thin film and surfaces research. The reflectometer operates in a monochromatic, angular dispersive mode collecting reflectivity data as a function of angle. The reflectometer operates with a vertical scattering plane, making it suitable for the study of air-solid, solid-liquid and air-liquid surfaces. Some of the representative data obtained on this instrument are shown below.

Figure 1. Reflectivity from a Bragg Mirror (nickel/titanium multilayer).

Figure 1 shows the reflectivity from a Bragg mirror composed of 25 bilayers of nickel and titanium deposited on float glass by Mirrotron Ltd (Budapest, Hungry) by RF magnetron sputtering. The structure of the Bragg mirror was modelled using the Parratt formalism with corrections for background and resolution smearing. The structure was based on 25 repeating bilayers, with the scattering length density (SLD) of the Ni and Ti layers set at their literature values and the layer thickness of each component refined. The roughness of the substrate as well as the interface between the Ni and Ti layers were also refined. The Ni layer was found to be 115 ? and the Ti layer was found to be 81?  thick, with an interfacial roughness of 7 ?.

Figure 2. Reflectivity profile from a plasma-coated silicon wafer, measured against D2O using a solid-liquid cell (courtesy of Patrick Hartley, Keith McLean, Sofia Oiseth, CSIRO Molecular Science, Clayton.)

Figure 2 shows the reflectivity profile from a plasma-coated silicon wafer, measured against D2O using a solid-liquid cell.

The plasma polymerised film of allylamine was deposited on a single crystal silicon wafer (<111>, 100 mm diameter x 10 mm thick) in a custom-built reactor at CSIRO Molecular Sciences.  Prior to deposition the wafers were cleaned in aqua regia and piranha solution before being thoroughly rinsed with MilliQ water and blown dry with nitrogen gas.

With this solid-liquid cell neutrons enter through the side of the silicon wafer, and are reflected from the wafer/polymer and polymer/D2O boundaries, before exiting through the other side of the wafer.  The structure of the deposited allylamine polymer was modelled using the Parratt formalism with corrections for background and resolution smearing.  The structure was modelled on a single polymer layer upon a native SiO2 layer with 5 ? gaussian roughness at each of the interfaces.  Figure 2 shows the observed (o) and refined (____) neutron reflectivity profiles. The native oxide layer was found to be 13 ?  thick, whilst the thickness of the allylamine plasma polymer layer was 273  ? with a refined SLD of 3.92x10-6 ?-2.

Previous data collected at the air-solid interface for this film, using the SURF reflectometer at the ISIS facility, were used to refine a film thickness of 252 ? with a SLD of 1.98x10-6 ?-2.  These previous data along with X-ray Photoelectron Spectroscopy measurements suggests a composition for this film of C79O9N12H68 (based on a density of 1 gcm-3).  The combination of the swelling of the film and the substantially larger SLD when in contact with D2O suggests that a significant amount of D2O (SLD = 6.37x10-6 ?-2) penetrates into the film.  Using the SLD of the films when measured against air and against D2O we estimate the relative amounts of D2O and polymer in the film to be 44 %v/v D2O and 56 %v/v plasma polymer.

Figure 3. Polymer spin coat layer on silicon with titania nanoclusters embedded in polymer matrix.  Courtesy of Nano-engineering, Functional Materials, Materials and Engineering Science, ANSTO.

Figure 3 shows the reflectivity from a silicon wafer coated with Polymethyl methacrylate (PMMA) containing 10 %wt of titania nanoclusters.  The polymer layer was spin-coated onto the silicon wafer from a dichloromethane solution.

The reflectivity was collected at a constant dQ over the Q range 0.008 to 0.12 ?-1.  The reflectivity was analysed using the Abeles optical matrix method (using the Drydoc program), with corrections for background and resolution smearing.

The structure was modelled on a single polymer layer upon a native SiO2 layer with 5 ? gaussian roughness at each of the interfaces.  Figure 3 shows the observed (o) and refined (____) neutron reflectivity profiles. The native oxide layer was found to be 13 ? thick, whilst the thickness of the polymer matrix was 585 ? with a refined SLD of 0.96-10-6 ?-2.

Figure 4. Reflectivity from hydrated PMMA layers deposited on an Si wafer, measured against two different mixtures of D2O/H2O.  Courtesy of Rossen Sedev, Ian Wark Institute, Adelaide.

Figure 4 shows the reflectivity from a silicon wafer coated with Polymethyl methacrylate (PMMA), measured against D2O/H2O using a solid-liquid cell.  The polymer layer was spun coated onto the silicon wafer from a toluene solution.  In order to help with unambiguous structure determination two different contrasts were measured (6.33x10-6 ?-2 and 3.78x10-6 ?-2).

The reflectivity was collected at constant dQ/Q over the Q range 0.01 to 0.2 ?-1.  The reflectivity was analysed using the Abeles optical matrix method (using the Wetdoc program), with corrections for background and resolution smearing.  A gaussian roughness of 5 ? was used at each of the interfaces.  Both the datasets were co-refined.

In order to fit the data satisfactorily it was necessary to divide the surface PMMA layer into two separate regions.  The first layer (closest to the native SiO2 layer, 10 ? thick) was 24 ? thick and was hydrated with 33 %v/v of water.  In contrast the outer layer, which was 61 ? thick and closest to water phase, was less hydrated, with only 5 %v/v water penetration. 

Thus, it seems that water penetration through the PMMA film results in a hydrated layer at the silicon surface.  However, this water penetration does not seem to hydrate the upper part significantly, which remains fairly "dry".  One possible reason for this behaviour may be void formation at the surface, since the film was not annealed.

Further Improvements
Important changes have been made to scan protocols, which have improved data collection rates and statistics.

Previous measurements (including those made above) were made by using constant slit openings over a given Q range.  The slits were then opened further for the next Q range.  Unfortunately, this meant that the neutron flux onto the sample was not optimised, since the footprint decreases as Q gets larger.

However, by changing the way that the control program is used data acquisition is now taking place with a constant sample footprint and constant resolution (dQ/Q).  This boosts the data collection rate and hence, the statistics. 

Figure 5. Detector Count rates for Quartz/D2O interface.  Constant footprint as well as varying underillumination.

Lastly, the introduction of a shielded flight tube that reduces neutronic background and the accurate measurement and subtraction of a background has enabled us to reach reflectivities approaching 10-6.  Figure 6 shows an example of background subtraction for quartz measured against D2O.

Figure 6. Reflectivity from a Quartz/ D2O interface with background subtracted.